Read Anytime Anywhere Easy Ebook Downloads at ebookmeta.

com

Machine Learning and Optimization for Engineering

Design 1st Edition Apoorva S. Shastri

https://ebookmeta.com/product/machine-learning-and-
optimization-for-engineering-design-1st-edition-apoorva-s-
shastri/

OR CLICK HERE

DOWLOAD EBOOK

Visit and Get More Ebook Downloads Instantly at https://ebookmeta.com

 Recommended digital products (PDF, EPUB, MOBI) that
you can download immediately if you are interested.

AI Metaheuristics for Information Security in Digital

Media 1st Edition Apoorva S. Shastri

https://ebookmeta.com/product/ai-metaheuristics-for-information-
security-in-digital-media-1st-edition-apoorva-s-shastri/

ebookmeta.com

Machine Learning-based Design and Optimization of High-

Speed Circuits 1st Edition Vazgen Melikyan

https://ebookmeta.com/product/machine-learning-based-design-and-
optimization-of-high-speed-circuits-1st-edition-vazgen-melikyan/

ebookmeta.com

Stochastic Optimization for Large-scale Machine Learning

1st Edition Chauhan

https://ebookmeta.com/product/stochastic-optimization-for-large-scale-
machine-learning-1st-edition-chauhan/

ebookmeta.com

The CELTA Course Trainee Book 2nd Edition Peter Watkins

https://ebookmeta.com/product/the-celta-course-trainee-book-2nd-
edition-peter-watkins/

ebookmeta.com
Beyond Earth : a chronicle of deep space exploration,
1958-2016 2nd Edition Asif A. Siddiqi

https://ebookmeta.com/product/beyond-earth-a-chronicle-of-deep-space-
exploration-1958-2016-2nd-edition-asif-a-siddiqi/

ebookmeta.com

Reckless Soul Dirty Souls MC 2 1st Edition Emma Creed

https://ebookmeta.com/product/reckless-soul-dirty-souls-mc-2-1st-
edition-emma-creed/

ebookmeta.com

The Matrix Model 4th Edition L Michael Hall

https://ebookmeta.com/product/the-matrix-model-4th-edition-l-michael-
hall/

ebookmeta.com

A Very English Murder - Eleanor Swift Cozy Mystery 1

Verity Bright Et El

https://ebookmeta.com/product/a-very-english-murder-eleanor-swift-
cozy-mystery-1-verity-bright-et-el/

ebookmeta.com

The Struggle for the Past How We Construct Social Memories

1st Edition Elizabeth Jelin

https://ebookmeta.com/product/the-struggle-for-the-past-how-we-
construct-social-memories-1st-edition-elizabeth-jelin/

ebookmeta.com
Teacher The Henrietta Mears Story 1st Edition Marcus
Brotherton.

https://ebookmeta.com/product/teacher-the-henrietta-mears-story-1st-
edition-marcus-brotherton/

ebookmeta.com
Engineering Optimization: Methods and Applications

Apoorva S. Shastri
Kailash Shaw
Mangal Singh Editors

Machine
Learning and
Optimization
for Engineering
Design
Engineering Optimization: Methods and
Applications

Series Editors
Anand J. Kulkarni, Institute of Artificial Intelligence, Dr. Vishwanath Karad MIT
World Peace University, Pune, Maharashtra, India
Amir H. Gandomi, Faculty of Engineering and Information Technology, University
of Technology Sydney, Sydney, NSW, Australia
Seyedali Mirjalili, Brisbane, QLD, Australia
Nikos D. Lagaros, National Technical University of Athens, Athens, Greece
Warren Liao, Department of Construction Management and Industrial Engineering,
Louisiana State University, Baton Rouge, LA, USA
Optimization carries great significance in both human affairs and the laws of nature.
It refers to a positive and intrinsically human concept of minimization or maxi-
mization to achieve the best or most favorable outcome from a given situation.
Besides, as the resources are becoming scarce there is a need to develop methods
and techniques which will make the systems extract maximum from minimum use
of these resources, i.e. maximum utilization of available resources with minimum
investment or cost of any kind. The resources could be any, such as land, mate-
rials, machines, personnel, skills, time, etc. The disciplines such as mechanical,
civil, electrical, chemical, computer engineering as well as the interdisciplinary
streams such as automobile, structural, biomedical, industrial, environmental engi-
neering, etc. involve in applying scientific approaches and techniques in designing
and developing efficient systems to get the optimum and desired output. The multi-
faceted processes involved are designing, manufacturing, operations, inspection and
testing, forecasting, scheduling, costing, networking, reliability enhancement, etc.
There are several deterministic and approximation-based optimization methods that
have been developed by the researchers, such as branch-and-bound techniques,
simplex methods, approximation and Artificial Intelligence-based methods such
as evolutionary methods, Swarm-based methods, physics-based methods, socio-
inspired methods, etc. The associated examples are Genetic Algorithms, Differen-
tial Evolution, Ant Colony Optimization, Particle Swarm Optimization, Artificial
Bee Colony, Grey Wolf Optimizer, Political Optimizer, Cohort Intelligence, League
Championship Algorithm, etc. These techniques have certain advantages and limi-
tations and their performance significantly varies when dealing with a certain class
of problems including continuous, discrete, and combinatorial domains, hard and
soft constrained problems, problems with static and dynamic in nature, optimal
control, and different types of linear and nonlinear problems, etc. There are several
problem-specific heuristic methods are also existing in the literature.
This series aims to provide a platform for a broad discussion on the devel-
opment of novel optimization methods, modifications over the existing methods
including hybridization of the existing methods as well as applying existing opti-
mization methods for solving a variety of problems from engineering streams.
This series publishes authored and edited books, monographs, and textbooks. The
series will serve as an authoritative source for a broad audience of individuals
involved in research and product development and will be of value to researchers and
advanced undergraduate and graduate students in engineering optimization methods
and associated applications.
Apoorva S. Shastri · Kailash Shaw · Mangal Singh
Editors

Machine Learning
and Optimization
for Engineering Design
Editors
Apoorva S. Shastri Kailash Shaw
Institute of Artificial Intelligence Department of Computer Science
Dr. Vishwanath Karad MIT World Peace and Engineering
University Symbiosis International University
Pune, Maharashtra, India (Deemed University)
Pune, Maharashtra, India
Mangal Singh
Department of Electronics
and Telecommunications
Symbiosis International University
(Deemed University)
Pune, Maharashtra, India

ISSN 2731-4049 ISSN 2731-4057 (electronic)

Engineering Optimization: Methods and Applications
ISBN 978-981-99-7455-9 ISBN 978-981-99-7456-6 (eBook)
https://doi.org/10.1007/978-981-99-7456-6

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature
Singapore Pte Ltd. 2023

This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether
the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse
of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by similar
or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors, and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721,
Singapore

Paper in this product is recyclable.

Preface

Machine learning is a data-driven approach to problem solving. It uses calcula-

tion tools to learn and identify patterns or relationships to predict results or to
make decisions resulting from the minimal or maximum chance of either error or
success, depending on the problem. Similarly, optimization algorithms are highly
efficient algorithms which focus on finding solutions to highly complex engineering
design problems like traveling salesman problems, scheduling problems, profit maxi-
mization, etc. This book aims to provide a collection of state-of-the-art scientific
and technical research papers related to machine learning based algorithms in the
field of optimization and engineering design. The theoretical and practical devel-
opment for numerous engineering applications such as smart home, ICT-based irri-
gation systems, academic success prediction, future agro-industry for crop produc-
tion, disease classification in plants, dental problems and solutions, loan eligibility
processing, etc., and their implementation with several case studies and literature
reviews are included as self-contained chapters. Additionally, the book intends to
highlight the importance of study and effectiveness in addressing the time and
space complexity of problems and enhancing accuracy, its analysis, and validations
for different practical applications by acknowledging the state-of-the-art literature
survey. The book targets a larger audience by exploring multidisciplinary research
directions such as computer vision, machine learning, artificial intelligence, modi-
fied/newly developed machine learning algorithms, etc. to enhance engineering
design applications for society. State-of-the-art research work with illustrations and
exercises along with pseudo-code has been provided here.
Chapter “A Short Review of Machine Learning Techniques for Thermal, Energy
and Electrical Engineering Applications” represents a brief review of machine
learning techniques for thermal, energy and electrical engineering applications. ML
models are widely used in engineering applications to solve complex problems
that existing technology cannot handle. Although LightGBM performs faster than
XGBoost, many research literatures focused on XGBoost base machine learning
model, which performed better with respect to prediction, classification and fore-
casting models. In this work, an evolutionary algorithm is used in a neural network
for feature handling. The prediction model has been used for both regression and

v
vi Preface

classification engineering applications. Anomaly detection is considered to detect

undesired work during the running of any machine component. Time series models
are used for forecasting models. This article demonstrates how a machine learning
model improves the performance of engineering applications in thermal engineering,
energy, and electrical appliances.
Chapter “Design of Intelligent ICT Irrigation System Using Crop Growth Big Data
Analysis” suggests and designs a customized, intelligent irrigation system algorithm
using big data analysis based on the growth of cultivated crops. In addition, an
irrigation system was designed according to temperature, humidity, piping, light
quantity, and water content of crops. Through this, the control monitoring based
on the recurrent neural network (RNN) was applied by utilizing big data analysis.
In this paper, we designed a solution that is easy to manage and easy to use for
cultivation and growth by using a customized, intelligent watering system and various
ICT sensors. In addition, through post-management, the system was designed to
provide easy usability to users with low technological acceptance by changing S/W
and major control devices according to changes in cultivated crops. Designed and
proposed in the paper, the customized and intelligent watering system can be used to
provide uncomplicated usage using initial modeling for crops. In addition, due to the
user-customized intelligent irrigation system, it is possible to maintain the facility
through simple monitoring through a system that continuously self-feedback and
decision-making even with a simple setup at the initial step of installation.
Chapter “OpenCV and MQTT Based Intelligent Management System” proposed
a management system which is intelligent and can perform identification, counting,
and calculation of density of vehicles. After calculating the traffic density, the system
classifies the density into low, medium, and high density with the help of a decision
algorithm. This system is based on Python programming, and the libraries used in
Python are Open-source Computer Vision, NumPy, chardet, and time library. The
system is implemented in IoT-based platform Message Queuing Telemetry Transport.
The system methodology is carried out in four steps. The first step is vehicle detection
and counting. In the second step, the number plate of the vehicle is detected as well
as displayed. The third step includes traffic density detection and finding out the
emergency vehicle based on GPS Tracking using ESP32 and IoT over MQTT. Last
step is related to pollution monitoring. This complete model for the system is made,
and the results are shown in the chapter.
Chapter “A Machine Learning Model for Student’s Academic Success Predic-
tion” presents the importance of student’s academic success prediction, since their
study and development are the major areas of interest. Higher education institutions
are frequently interested in learning about students’ success rates during their studies.
As a result, they must employ a variety of strategies for predicting student success,
including physical assessment, statistical methodologies, and modern data mining
techniques. Educational Data Mining facilitates the improvement of teaching and
learning processes. The goal of the current study is to create data mining models
for forecasting student performance based on personal, pre-university, and post-
university factors. The dataset used in this study contains information on Undergrad-
uate students pursuing a B.Sc. of three years with 434 instances in the dataset, each
Preface vii

with 41 attributes. Attributes are classified as academic and non-academic followed

by exploratory data analysis. The dataset is subjected to decision trees and multilayer
perceptron algorithms with regression analysis. Precision recall values are calculated
using different attributes’ sets. Testing accuracy of j48 algorithm with all attributes
is maximum which is greater than 90%.
Chapter “Intelligent Agro-Industry for Crop Production Considering Soil Prop-
erties and Climatic Variables to Boost Its Efficiency” evaluates the need to develop a
system that boosts the efficiency of Agro-industry for crop production. Making sure
that everyone in this country gets to eat nutrient rich food in a feasible way is the
objective of all people who are associated with the agricultural industry. India being
an agricultural country, holds the second position in having arable land resources
with 20 various agricultural climatic regions and 46 of the 60 different types of
soil found in the world. But the unfamiliarity of farmers with the climatic condition
and the soil type in a certain region and the denial by farmers to adopt a newer
approach for growing crops is making production slower and the soil more vulner-
able to losing its fertility over a very short period of time and hence lesser yield. The
goal of researchers is to make crop production easier, simpler, faster, and more cost-
effective. For the above-mentioned goal, researchers adopt machine learning models.
In the proposed chapter based on the climatic condition and soil nutrients, the produc-
tion of suitable crops is determined using the supervised and unsupervised machine
learning algorithms. Logistic Regression (LR), Support Vector Machine (SVM),
Decision Tree, and K-nearest neighbor machine learning algorithms are used under
supervised learning and K-Means clustering is used under unsupervised learning.
The outcome of the machine learning algorithms is evaluated in terms of accuracy,
precision, recall, F-1 score, and confusion matrix.
Chapter “Disease Classification in Cassava Plant by Artificial Neural Network”
addresses the details and development of a machine learning model for detecting
cassava plant leaf disease. The cassava plant is the world’s second-largest producer
of carbohydrates and is grown to provide a large source of nutrition for people in
tropical climates. It is particularly susceptible to bacterial, viral, and fungal infections,
which restrict plant development and, as a result, decrease productivity. An Artificial
Neural Network (ANN) model is implemented which would classify and differentiate
the images into four diseased classes and one healthy class. The dataset utilized in
this study is drawn from the Kaggle challenge 2020 which includes photos of cassava
plant leaves from four disease classifications that affect cassava. The outcomes show
that the total accuracy of the training data generated was 61.45%, and the highest
accuracy of the test data generated from this ANN model was 61.88%.
Chapter “Exploring the Synergies: A Comprehensive Survey of Blockchain Inte-
gration with Artificial Intelligence, Machine Learning, and IoT for Diverse Applica-
tions” illustrates the applications of Blockchain technology, which is a decentralized
system for confirming transaction validation that is safe, verifiable, and relevant to
several sectors, including finance, the Internet of Things, big data, cloud computing,
and edge computing. Combining Blockchain, AI, ML, and IoT has the obvious benefit
of granting AI greater autonomy and legitimacy. This chapter presents a thorough
viii Preface

and three-dimensional analysis of the integration of blockchain with Artificial Intel-

ligence (AI), Machine Learning (ML), and the Internet of Things (IoT) describing
potential benefits and challenges. This chapter presents current issues and challenges
in the field as well as prospective directions for future study into how to combine
blockchain with AI, ML, and IoT.
Chapter “Loan Eligibility Verification by Using Ensemble ML Techniques”
presents a novel machine learning technique for Loan eligibility verification. Nowa-
days, people applying for loans has become very common, many people apply for
loan approval daily. However, all the applications cannot get approval. We witness
several cases every year where applicants take a large amount of loan and don’t pay
it back to the banks: thus, leading to monetary loss for banks and financial institutes.
Hence, banks or financial institutes have a high risk associated with this decision.
And the main objective of this decision-making is to grant the loan to deserving
applicants, who will repay the loan without any difficulties. After reviewing many
prediction-related classification-based research papers it has been found that in the
case of traditional machine learning (ML) algorithms like Decision Tree and Random
Forest algorithms give prodigious accuracy as compared to other ML algorithms.
Also, if we use further ensemble ML techniques and deep learning techniques then
we can achieve higher accuracy as well. In this study, we have collected loan data
from multiple data sources and built a model by using various ML algorithms, for
efficient loan approval prediction for a specific application. In this study, we have
deeply analyzed the bank credit risk by using ML algorithms like Logistic Regres-
sion (LR), Support Vector Machine (SVM), Random Forest (RF), XGBoost (XGB),
AdaBoost and k-nearest neighbor (KNN) on the data that determines the creditwor-
thiness of aspirants. We found that the XGB ensemble classifier shows remarkable
accuracy.
Chapter “An Intelligent System for the Classification of Dental Problems Based
on Bayesian Regularization Neural Network” addresses the need of Neural Network
based techniques for dental problems. In this chapter, an intelligent system which
can detect and classify dental problems as well as give consultation according to
the problem is proposed. This intelligent system is very useful in a scenario of
COVID-19 outbreak where doctor-patient contact should be minimal. The trained
neural network is the brain of the system. Bayesian regularization is used to train the
network. The learning algorithm is chosen after the comparison with other six back-
propagation algorithms on the basis of accuracy. The inputs to the neural network are
the various features extracted using image processing. Energy, correlation, entropy,
contrast etc. are the various features used to train the network. The images used in
image processing are of the four dental problems viz. mild fluorosis, severe fluorosis,
tooth decay and enamel hypoplasia. The images of healthy teeth are considered for
the comparison. MATLAB is used for the image processing and training of the neural
network. The trained network is linked with the MATLAB graphical user interface.
The MATLAB GUI displays the detected teeth problem and the consultation as the
patient’s teeth image is fed to it. The images of the teeth problems fed to the proposed
system and the results obtained are presented.
Preface ix

Chapter “Development of Smart Home System Based on EEG” aims to implement

a smart home system that can control various devices and home appliances in the
home for the elderly and people with disabilities. The developed system classifies
brain waves when a user thinks to the right or left and opens and closes the window
in the desired direction. To this end, a mobile application was developed to induce
direction-related brain waves from users, and a deep learning model for classifying
brain waves was proposed. In addition, a control program was developed to receive
the classified results and operate the motor responsible for window operation. This
system is an information and communications technology (ICT) convergence system
that combines brain computer interface (BCI) technology and internet of things
(IoT) technology to support convenient services at home. BCI technology was used
to emphasize user convenience for the elderly and the disabled, and deep learning
technology was used to increase the accuracy of classification of brain signals. The
developed system can be applied not only to Windows but also to various devices
and home appliances by adding a menu selection function of the mobile application.

Pune, India Apoorva S. Shastri

Kailash Shaw
Mangal Singh
Contents

A Short Review of Machine Learning Techniques for Thermal,

Energy and Electrical Engineering Applications . . . . . . . . . . . . . . . . . . . . . . 1
Nihar Ranjan Swain, Shreejith Pillai, and Sanket Kumar Yadav
Design of Intelligent ICT Irrigation System Using Crop Growth
Big Data Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Ssang-Hee Seo and Bong-Hyun Kim
OpenCV and MQTT Based Intelligent Management System . . . . . . . . . . . 25
Anand Mahajan, Satej Gadekar, Sumit Sagave, Smita Paithankar,
Bhushan Rakhonde, and Chetan Khadse
A Machine Learning Model for Student’s Academic Success
Prediction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Antara Asthana, Vidya Kumbhar, Prafulla B. Bafna, and Alka Gadakh
Intelligent Agro-Industry for Crop Production Considering Soil
Properties and Climatic Variables to Boost Its Efficiency . . . . . . . . . . . . . . 57
Manorath Jaydev, Ritika Kar, Biswa Ranjan Senapati,
and Subrat Kumar Nayak
Disease Classification in Cassava Plant by Artificial Neural Network . . . 75
Mandar Sapre, Vijaykumar S. Jatti, Pranjal Tiwari,
Niranjan Kodachakki, and Atharva Undale
Exploring the Synergies: A Comprehensive Survey of Blockchain
Integration with Artificial Intelligence, Machine Learning, and IoT
for Diverse Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Namita Kalyan Shinde, Ashutosh Seth, and Payal Kadam
Loan Eligibility Verification by Using Ensemble ML Techniques . . . . . . . 121
Sachin Bhoite, Surabhi Thatte, Ajit More, and Darshan Ruikar

xi
xii Contents

An Intelligent System for the Classification of Dental Problems

Based on Bayesian Regularization Neural Network . . . . . . . . . . . . . . . . . . . 135
Chetan B. Khadse, Yogesh Tatte, and Swati Jaiswal
Development of Smart Home System Based on EEG . . . . . . . . . . . . . . . . . . 153
Bong-Hyun Kim and Ssang-Hee Seo
About the Editors

Apoorva S. Shastri holds a Ph.D. in Optimization Algorithms and Applications from

Symbiosis International (Deemed University), a Master of Technology (M.Tech.) in
VLSI Design, and a Bachelor of Engineering in Electronics and Product Design
Technology from R.T.M.N.U, Nagpur. She has also earned a Diploma from the
Government Polytechnic, Nagpur. She worked as a guest faculty at the Centre for
Development of Advanced Computing (C-DAC), Pune. Currently, she is a research
assistant professor at the Institute of Artificial Intelligence at MITWPU, Pune, India.
Her research interests include optimization algorithms, VLSI design, multi-objective
optimization, continuous, discrete, combinatorial optimization, complex systems,
manufacturing, and self-organizing systems. She developed socio-inspired optimiza-
tion methodologies such as Multi-cohort Intelligence Algorithm and Expectation
Algorithm. She has published several research papers in peer-reviewed journals,
chapters, and conferences along with one Springer authored book.

Kailash Shaw is an associate professor in the Department of Computer Science and

Engineering, Symbiosis Institute of Technology, Pune, India, and also working with
the Department of Artificial Intelligence and Machine Learning (AIML). Adminis-
tratively, he is a member of the Board of Studies (AIML). His research and teaching
interests are in the domain of data science, programming skills, optimization engi-
neering, bio-inspired algorithms, and ICT-based education. In terms of teaching, he
is the recipient of the Visweswarya Prativa Puraskar in recognition of his outstanding
performance in research. He supervised over fifteen master’s dissertations. In total,
he has authored over 50 peer-reviewed articles and edited two research books
proceeding. He has served as a convener and organized two international confer-
ences whose proceedings are indexed in Springer. He has also served as the track
editor and the session chair for many international conferences.

Mangal Singh is working as an associate professor in Electronics and telecommu-

nication Engineering at Symbiosis Institute of Technology, Symbiosis International
(Deemed University), Pune. He has an experience of more than 21 years in the field of
teaching, research, and administration. He obtained his graduation in Electronics and

xiii
xiv About the Editors

Telecommunication Engineering from the National Institute of Technology (formally

known as GEC), Raipur, Chhattisgarh, and his M.Tech. in Communication Engi-
neering at Jadavpur University, Kolkata, West Bengal. He obtained his Ph.D. in
Communication Engineering from the National Institute of Technology, Rourkela,
Odisha. He has served as an associate professor of electronics and communica-
tion Engineering at the Institute of Technology, Nirma University, Ahmedabad, and
an associate professor of electronics and communication Engineering at Chhatra-
pati Shivaji Institute of Technology, Durg, Chhattisgarh. He has three Indian patents
published in his credit. He has guided several PG dissertations. He is a senior member
of IEEE and a life member of the IETE and ISTE, India.
A Short Review of Machine Learning
Techniques for Thermal, Energy
and Electrical Engineering Applications

Nihar Ranjan Swain, Shreejith Pillai, and Sanket Kumar Yadav

Abstract Machine Learning (ML) is the most rapidly evolving technology for
problem solving in engineering. It helps intelligence systems learn each typical
pattern of the environment by creating submodules from historical data that take
predictions or cluster models for new inputs. ML models are widely used in engi-
neering applications to solve complex problems that existing technology cannot
handle. Although LightGBM performs faster than XGBoost, more research found
XGBoost base machine learning model performed best in prediction, classification
and forecasting models. For feature handling, an evolutionary algorithm is used in
a neural network. The prediction model is used for both regression and classifica-
tion engineering applications. Anomaly detection is used to detect undesired work
during running of any machine component. Time series models are used for fore-
casting model. This article demonstrates how a machine learning model improves
the performance of engineering applications in thermal engineering, energy, and
electrical appliances.

Keywords Machine learning · Engineering design · Data representation ·

Thermal engineering · Electrical engineering · Energy · Optimization of designing

1 Introduction

ML is a branch of AI that creates models and algorithms for computers to learn from
historical data of appliances and make decisions based on these data patterns. These
models are trained using large datasets, enabling them to automatically spot patterns,
linkages, and insights that human programmers might miss. Machine learning has
become significantly more relevant and useful in the engineering area. Engineers
are given strong tools to analyse intricate systems, enhance designs, and enhance

N. R. Swain (B) · S. Pillai · S. K. Yadav

Institute of Artificial Intelligence, Dr Vishwanath Karad MIT World Peace University, 124 Paud
Road, Kothrud, Pune 411038, Maharashtra, India
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 1
A. S. Shastri et al. (eds.), Machine Learning and Optimization for Engineering Design,
Engineering Optimization: Methods and Applications,
https://doi.org/10.1007/978-981-99-7456-6_1
2 N. R. Swain et al.

decision-making. Engineers may get information from enormous amounts of data

by utilising machine learning techniques, which will improve efficiency, accuracy,
and innovation across a range of engineering applications. With advancement of ML
application in software industry will solve more engineering problems like large
language model of ChatGPT is able to answer any type of text data based on the
number of trained ML models. With the application of AIML base intelligence system
able to solve all types of engineering problems, the advancement of ML algorithms
results in more simplified solutions to make life easier [1]. Machine Learning refers
to statistical modelling methods that have recently generated a lot of interest in the
field of software and services industry with other engineering problem solving. Arthur
Samuel coined the term “machine learning” in 1959. Machin learning is classified into
three types: supervised, unsupervised, and reinforcement learning. The advancement
of computer hardware systems such as graphics processing units (GPU) and tensor
processing units (TPU) will result in faster computation for machine learning models
[2]. Reinforcement learning is used to control agents or any robotic application.
However, development of deep learning to remember the sequence of policy on
history of data or optimize data from heuristic method gives a power boost to train
such environment, such an approach is called Deep Q-Network (DQN) [3]. This
research work will show how machine learning can be opted to solve different types
of engineering problems.

2 Intuition About Machine Learning and Optimization

2.1 Types of Data in Machine Learning Model

Machine learning assists in learning the pattern of any object by knowing its history
and making a future prediction or categorization. As a result, machine learning is
completely data dependent; rather than a new object or problem statement, it fails to
identify without prior data. Everything in the world is stored as data. The detailed
taxonomy of data is shown in Fig. 1. Data is broadly classified as structured and
unstructured. Structured data follow a specific pattern, such as the number of rows
and columns, for example, tabulated results to find torque of IC engine testing with
other setting parameters. Unstructured data which have not exact pattern such as
image, video and audio data. Structured data is classified into two types: quantitative
and qualitative. Quantitative data are those that can be measured, such as a person’s
weight. There are two types of quantitative data: continuous and discrete. A contin-
uous data set is any quantitative data that does not have any constraints while those
data that repeats with a fixed value is referred to as discrete value. A qualitative data
set that provides some value or service in terms of time. The two types of qualitative
are nominal (no order or sequence) and ordinal (with an order). Before preparing a
machine learning model, unstructured data such as audio, video, and text may require
Big Data analysis. Before developing any machine learning model, all data should
A Short Review of Machine Learning Techniques for Thermal, Energy … 3

Types of Data

Structured Unstructured data

Quantitative Qualitative/Categorical
Audio Video Text

Continuous Discrete Nominal Ordinal

Fig. 1 Types of data

be converted into numerical form. The most common data collection methods are
database, IOT base, cloud base, and cloud base [2, 4, 5].

2.2 Data Pre-Processing

2.2.1 Structural Data Pre-processing

Dropping duplicate data from the initial stage of structural data pre-processing
reduces data bias. Feature Engineering encompasses all data pre-processing
processes. Feature engineering is broadly divided into feature transformation, feature
construction, feature selection, and feature extraction. All data points are converted
into numerical values using feature transformation. Missing value handling is
included in the feature transformation steps; if 70–80% of the data is missing for
any feature, we drop it in the worst case also known as complete case analysis. Cate-
gorical missing variables are replaced with the most frequent value (mode) or, for
simplicity, the value ‘missing’ is replaced (which will be separately handled). Impu-
tation of mean, median, mode, random variable, and end of distribution (robustness
of outlier) is used to fill quantitative univariate missing data. KNN Imputer and iter-
ative method are used to multivariate quantitative missing value. Categorical value
conversion includes one hot encoding, whereas the label encoder is only used for
the target variable. Outliers in the data can influence any machine learning model’s
prediction, so outlier data is compressed into two boundary layers to avoid robustness.
Feature scaling is a technique for reducing all independent features to a manageable
range. If necessary, the number of input variables is very small. Feature construction
aids in the creation of new features from given data using domain knowledge of the
input data. Feature selection technique selects the best suitable variables for machine
learning model by using Correlation, Variance Influence Factor (VIF), Chi-square or
4 N. R. Swain et al.

ANOVA test. To extract all data without deleting features, feature extraction tech-
niques such as Principal component analysis (PCA), Independent component anal-
ysis (ICA), Linear Discriminant Analysis (LDA), Locally linear embedding (LLE),
t-distributed Stochastic Neighbor Embedding (t-SNE), and deep learning models
such as Autoencoder are used [6–12].

2.2.2 Text Data Pre-processing

Natural language processing (NLP) includes text data pre-processing. Text data is
not the same as structural data. Text data refers to any type of written language
text. Knowledge of a specific language’s written style, grammar, and syntax aids
in understanding the in-depth pattern of text. Expand contraction and remove the
contraction from the sentence (I’ll to I will). Other proceeding steps are converting
all text into lower case, removing punctuations, and removing stopwords. If the input
text contains a small number of words, text augmentation will aid in the develop-
ment of a perfect machine learning model. Text augmentation includes paragraph
shuffling, word replacement (replacing with synonyms), syntax tree manipulation
(paraphrasing), random word deletion, and random word deletion. Tokenization is
the process of converting a sentence into a token form. When working with NLP, the
meaning of the word is more important than the word itself. Stemming will remove
affixes from each word to convert it to its base form; stemming is faster, but it can
sometimes convert a word to its base form with no meaning. Lemmatization is similar
to stemming in that it converts meaningful base forms of words, but it is slower.
Parts of speech (without removing stopwords) will aid in the schematic meaning of
a sentence, such as noun, verb, adverb, and so on. Vectorization is a technique used
to convert text into vector form using Onehot encoding, Bag of words, N-grams,
and TF-IDF. Text data machine learning applications include rephrasing, sentiment
analysis, chatbots, language translation, OCR, and Large Language Models [13, 14].

2.2.3 Audio Data Pre-processing

Audio data pre-processing includes audio augmentation if the durability of an audio

is very low or if the pattern to identify the speech is not perfect. Audio augmentation
includes noise injection, which involves inserting random noise into audio, shifting
audio to the left or right, or fast forwarding, changing audio speed, and changing
audio pitch. Then convert audio to text and proceed with text data pre-processing
as mentioned above. speech recognition, voice translation, and many others are the
application audio data ML model [15].
A Short Review of Machine Learning Techniques for Thermal, Energy … 5

2.2.4 Video Data Pre-processing (Computer Vision)

A video dataset is a collection of images that are displayed frame by frame. When
a sufficient number of images cannot be found, to make the best machine learning
model image augmentation will generate a number of images by using geometric
transformations such as flipping, cropping, stretching, or zooming, colour space
transformation converting image brightness, contrast, and colour channel, and so on.
Kernel filters are used to blur or sharpen images, randomly erase parts of images, and
mix input image. Pixel normalization, Pixel standardization, and Pixel centring are
all examples of image scaling. Application of video data machine learning projects
are object classification, object detection, object tracking, etc. [16].

2.3 Types of Machine Learning Algorithm

Machine learning is divided into four categories: supervised, unsupervised, semi-

supervised, and reinforcement learning. From given datasets, supervised learning
predicts both classification and regression values. An unsupervised algorithm is used
to cluster and associate one item or feature with other items or features. The semi-
supervised algorithm aids in the annotation of datapoints. Reinforcement learning is
used to control any system in response to environmental data. Figure 2 depicts the
entire machine learning taxonomy.

2.3.1 Supervised Learning

Supervised machine learning algorithms are those that can predict the target variable
based on an independent feature. The supervision algorithm incorporates both a
classification and a regression model. Figure 3 depicts the structure of the simplex

Types of Machine
Learning

Supervised Unsupervised Semi-Supervised Reinforcement

Learning Learning Learning Learning

Classification Regression Clustering Association Classification Clustering Classification Control

Fig. 2 Types of machine learning

6 N. R. Swain et al.

supervise algorithm. The solution to a regression problem is to solve the target value
with a continuous value, such as ambient temperature measurement. A classification
problem is used to determine to which class does the input data belong, such as
email spam classification. The target value is used to classify which class input data
belongs to, such as email spam classification. Again, classification is divided into
two types: binary classification with two variables and multiclass classification with
more than two variables.
This section goes over the entire structure of a basic machine learning algorithm
that is used in engineering. Linear Regression is used to predict linear datasets by
constructing a prediction line from the coefficient of variable and the error term. To

Supervised Learning

Regression Classification

1. Linear Regression 1. Logistic Regression

1. Simple Linear Regression

2. Decision Tree
2.Multiple Linear Regression
3. Support Vector Machine
3.Polynomial Regression

4.Regularization 4. K Nearest Neighbor

5.Elastic Net Regression

5. Naïve Bayes

2. Decision Tree
6. Ensemble
3. Support Vector Machine
Bagging
4. K Nearest Neighbor
Random Forest
5. Bayesian Regression

6. Ensemble Boosting

Bagging Ada Boost

Random Forest Gradient Boost

XGBoost
Boosting
LightGBM
Ada Boost
CatBoost
Gradient Boost

XGBoost

LightGBM

CatBoost

Fig. 3 Supervised machine learning algorithm

A Short Review of Machine Learning Techniques for Thermal, Energy … 7

optimize these parameters, the gradient descent method is used. When error terms are
cautiously reduced, the model overfits (ML model gives the best output for train data
while it performs worst in test or validation data). By introducing weight multiplica-
tion in the gradient of the loss function, regularization is used to overcome overfit-
ting (L1 and L2 regularization are widely used). The loss function (Error) represents
the overall difference between the actual and predicted values. Common loss func-
tions are mean absolute error, mean square error, root mean square error. Regression
model validation R2 score and adjusted R2 score (importance of new data to predic-
tion model) are employed. Logistic Regression is a classification algorithm used to
predict class. The model must be fine tuned to achieve the best coefficient matrix
based on the input feature, then the multiplication factor with the sigmoid function
compresses the range 0–1, and the threshold (through AUC-ROC curve) will identify
between two classes. SoftMax regression is used for multiclass classification using
the SoftMax function. The validation matrices for each classification problem are
accuracy, precision, recall, and F1 score, while other parameters such as specificity
and sensitivity model prediction rate are based on the actual class. Decision Tree (DT)
is a tree search algorithm that splits all the data and predicts all the data according to
belonging feature, using different criteria, a decision tree is used for both regression
and classification (Gini impurity, Entropy for classification model and MSE, MAE
for Regression model). Pruning the decision tree model aids in overcoming overfit-
ting. Support vector Machine (SVM) is a vector-based algorithm that can predict the
output of both classification and regression problems. The complexity of a feature
can be extracted using a kernel trick (linear kernel, polynomial kernel, Gaussian
kernel, Exponential kernel, Laplacian kernel, Sigmoid kernel, ANOVA radial basis
kernel). K Nearest Neighbor (KNN) is a distance matrix (Manhattan, Euclidian,
Minkowski, etc.) method for regression or classification models that predict new
variables using nearby features. Naive Bayes is a conditional probability method
that is used to create classification models. Ensemble algorithms such as bagging,
boosting, voting, and stacking will be used with greater accuracy to learn the in-depth
structure of data. Bagging employs a large number of subsamples to learn each algo-
rithm, and the combined results predict the model, such as Random Forest (RF).
Boosting improves the submodules by lowering the error rate of each model such
as (Adaboost, Gradient boosting, Extreme gradient boosting, CatBoost, and Light
gradient boost). Voting will combine the results of all subsample models, whereas
stacking is a sequence modelling method that employs other subsample modules.
To handle time series data separate supervised machine learning algorithms such as
autoregressive integrated moving average (ARIMA), Seasonal Autoregressive Inte-
grated Moving Average (SARIMA) and sequential data deep learning algorithm
[2].

2.3.2 Deep Learning

Deep learning is a type of supervised learning that can predict classification and
regression problems. It excels in large datasets, but unlike simple machine learning
8 N. R. Swain et al.

models, it cannot interpret data after it has been trained due to its black box approach.
The details taxonomy of the Deep learning model is shown in Fig. 4. Every deep
learning model’s basic architecture is a neural network. A neural network is a multi-
layer model that has a certain number of nodes in each layer. Multi-layer signifies
input layer is the shape of features, hidden layer captures deep intricate pattern of
data and output layer predicts the output. In a neural network model weights and
bias initialization is accomplished through the use of random non-zero weight and
bias initialization or the standard form weight initialization method, with the goal of
achieving optimal weights and biases. Summation of the specified weights are multi-
plied by the input feature and bias term for that node pass through activation function
(sigmoid, tanh, relu, leaky relu, and so on) to squeeze into a range. The following
layer applies the same principle with the previous layer’s input. This is known as
forward propagation. Backpropagation with appropriate error metrics such as MSE,
MAE, RMSE for regression, Binary cross entropy for two-class classification and
Categorical cross entropy, sparse categorical cross entropy for multiclass classifi-
cation model, followed by optimizer methods such as Rmsprop, gradient descent,
AdaDelta, Adam, Adagrad, and so on to finetune weights and biases. Deep neural
network with a large number of hidden layer although maximum model score always
tends to overfit, so to overcome the dropout layer, regularization and batch normal-
ization technique is used. During the training process with a large number of epochs,
problems such as vanishing gradient (very small weights update) and exploding
gradient with height gradient (weights jump in large steps) will be handled by fine
tuning the model. When the weights are not updated for very long epochs, the callback
method is used. Convolutional neural networks are used to capture image data edges
and classify them using a neural network, and the combination of all edges gives
complete image information. Convolution operation with tuned kernel size weights
reduces the shape of the image, and pooling and flattening extract the required feature.
Those features with labelled regression values will use IOU (intersection over union)
values to localise the required part of the image, while those with labelled classified
values will predict the class it belongs to. Recurrent neural networks (RNNs) are
most commonly used for sequential data such as text and time series data. Because
the RNN model is a short-term memory algorithm extension model such as the
gated Recurrent Unit (GRU) and the Long Short-Term Memory Network (LSTM)
aid in the long-term memory of data sequences. To some extent, the extension model
captures the text sequence; the encoder model aids in the handling of such files;
BERT and Large Language models are examples of encoder models. The Generative
Adversarial Network (GAN) is used to generate new data from input data with the
appropriate label. GAN is used to remove blurry images and generate new images
[3].

2.3.3 Unsupervised Learning

Unsupervised Learning is a technique for clustering or associating data with the

closest datapoints. Unsupervised learning datasets have only one independent feature.
A Short Review of Machine Learning Techniques for Thermal, Energy … 9

Deep Learning Algorithm

ANN-Artificial Neural Network

CNN-Convolutional Neural Network

GRU – Gated
RNN-Recurrent Neural Network
Recurrent unit

LSTM – Long
Encoder
Short-Term
Memory
GAN-Generative Adversarial Network

Fig. 4 Types of deep learning models

The clustering model is used to find the number of clusters to form the datasets. The
clustering models are K-means, Hierarchical and Density-based spatial clustering of
applications with noise (DBSCAN). K-means is a distance matrix method to find
the number of clusters, while the Elbow curve (using within-cluster sum of squared
distance) is used to determine the maximum number of clusters. Using inter- and
intra-cluster distance, the silhouette and Davis-Bouldin scores are used to validate
cluster size. To cluster with complex data patterns, hierarchical (Agglomerative and
divisive) and DBSCAN clustering are used. Association rule (Apriori model) is used
to associate one feature with another feature using association rule [17].

2.3.4 Semi-supervised Learning

Semi-supervised models are employed when a dataset contains both labelled and
unlabelled data. The semi-supervised model establishes a link between supervised
and unsupervised learning. Semi-supervised algorithms are self-training; a super-
vised model is built with small label data, whereas a pseudo label model is built with
large unlabelled data combing two models will be performed for the new dataset.
Co-training is another semi-supervised learning algorithm that is based on two-view
classifier data, where the trained model interprets the view of the classified label with
input features accurately [18].

2.3.5 Reinforcement Learning

Reinforcement learning (RL) is a type of machine learning that does not require
any input data, similar to supervised and unsupervised algorithms and it is used to
control the Intelligence system in the best possible way. The Agent (Intelligence
Engine) perceives data from the environment and takes action to the next state based
10 N. R. Swain et al.

on the reward function. RL consists of two types of mechanisms; a model-based

algorithm that partially understands the environment and uses an algorithm to find
the best possible reward function for the agent’s next steps. Model-free algorithms
learn from the unknowing environment with trial-and-error method to find the best
possible model. The agent takes an action (movement of agent based on problem
statement) based policy set to it. The policy is of two types; Deterministic policies
allow an agent to take only one action to reach a goal, whereas stochastic policies
allow an agent to take n actions to reach a goal [3].

3 Application of Machine Learning Engineering

Application

3.1 ML in Thermal Engineering

Monte Carlo Tree search machine learning algorithm detects four-band, wide-angle
nonreciprocal thermal radiation of 32 layers of SiO2 and InAs magneto-optical
material with Bayesian optimization, whereas previous studies detected single or
dual bands with limited angle [19]. As the demand for ionic liquids in industrial
applications grows, a predictive model is used for the surface tension of ionic fluid.
As the number of input features is large, optimization algorithms such as Genetic
Algorithms (GA), Particle Swarm Optimization (PSO), and Fruit-fly Optimization
algorithm (FOA) are used to best feature selection. Finally, all optimization-based
feature selection models combined with KNN are used to predict surface tension. A
combination of Genetic Algorithm for feature selection and KNN prediction model
yields a higher R2 score with a low prediction error [20]. After administering local
anaesthesia, medical thermal foot images are classified based on the region of interest
using supervised ML algorithms such as ANN, KNN, Random Forest, and SVM,
with ANN providing the highest accuracy and AUC for classifying images based
on time [21]. To maximize thermal efficiency, GA is used to optimize the cooling
system parameters of a vehicle engine. The thermal efficiency of a vehicle engine
was predicted using a neural network-based prediction model [22]. Electric cars
have grown in popularity in recent years; however, they have several drawbacks,
including a lengthy charging time, being a poor choice for long-distance travel, and
a Battery Thermal Management System (BTMS). As a workaround, lithium-ion
batteries have a high energy impact and can be used to power a vehicle. To maxi-
mize a vehicle’s range, state-of-the-art BTMS should perform optimally. Despite
the fact that PCM-graphite is the best performing battery cooling system in terms
of cooling, it is not optimal. To optimize cooling design, predictive and forecast
machine learning algorithms are used. The best ML algorithm for such a design is
Extreme Gradient Boost, which is used to predict the number of temperatures at a
given time. ARIMA is used to predict one temperature from another using an image
processing algorithm, whereas Support vector machining with RBF Kernel is used
A Short Review of Machine Learning Techniques for Thermal, Energy … 11

to predict one temperature from another [23]. The autonomous calibration method
for nuclear power plants is aided by machine learning models. The performance of
digital twins should match both simulation data and real time data in order to observe
the condition of the nuclear power plant as a source of electrical energy. Because
the data is continuous, subsample Neural network models are created to avoid the
repetition of data. Anomaly detection models are used to detect abnormal data; if
a sensor fails, the missing value imputation model takes over. Clustering of digital
twin and real used for model improvisation [24].

3.2 ML Application in Energy Sources

Wind energy conversion systems capture wind energy and convert it to electrical
power. Wind turbines (WT) are available in a variety of topologies and architectures.
It can vary depending on the axis orientation, hub, rigidity, number of blades, power,
and yaw control, as well as the position and speed of the rotor. Some of the difficult
problems that have been solved include wind power prediction, optimized control
of wind turbine generators, defect detection, predictive maintenance, and condition
monitoring. Machine Learning replaces a broad approach to controlling the system
through intelligence. ANN models were developed to forecast wind speed using
geometry and material inputs. To determine the best designs, a combination of ANN
models and the Multi-objective GA were used for optimization. The SVR method
was used to improve wind turbine monitoring indicators. Predictive control of wind
turbine generators using a combination of multi-objective particle swarm optimiza-
tion and machine learning methods. An ML-based day-ahead wind power forecasting
approach was developed using Multiple Linear Regression, DT, and RF. Developed
a model for predicting wind energy generation using fuzzy rules and ANN. SVR
and DT are used to forecast wind power. Wind generation forecasts could benefit
from the use of AdaBoost and XGBoost ensembles [25, 26]. Nanofluids (NFs) were
created as a result of efforts to improve heat transmission in solar collector systems.
Among other heat transfer properties, nanofluids have increased thermal conductivity
and specific heat capacity. Specific heat capacity aids in the monitoring of efficient
thermal energy management systems in order to maximize energy savings and mini-
mize system energy loss. The XGBoost model outperformed the Support vector
machine and Gaussian process regression in the specific heat capacity evaluation of
nanofluids for solar energy applications [27]. The electrical performance of a solar
chimney power plant (SCPP) with a photovoltaic thermal system (PVT) integrated
into it is evaluated. The Support Vector Regression (SVR) method is used with linear,
polynomial, Gaussian, and hybrid kernels to forecast and optimize system perfor-
mance under various operating conditions. To produce the most accurate results, the
global optimum points of the SCPP-PVT system are determined using the multi-
objective grasshopper optimization method, and the hyperparameters of the machine
learning models are optimized using the Grey Wolf Optimizer [28]. Meteorological
parameters such as rainfall, solar radiation, temperature, and snowpack level all have
12 N. R. Swain et al.

an impact on hydropower production. For hydropower prediction from metrological

data, ANN without feature reduction produced a better predictive model than SVM
[29].

3.3 Application of ML in Electrical Appliances

Detecting a person’s absence based on electrical appliances used could be useful in

determining whether the residents are present in their house. During the pandemic,
elderly people and home quarantined patients were among the scenarios considered.
On UK Domestic Appliance-Level Electricity (UK-DALE) datasets, C4.5 trees (a
variant of the decision tree algorithm) DT, RF, Naive Bayes (NB), Multilayer Percep-
tron (MLP), and Deep Neural Networks (DNN) [30]. Theft events such as consider-
able reduction of electricity consumption, electricity consumption dropping to zero at
random, consumption value is multiplied by a random number, generating a random
fraction of the mean consumption, reporting the mean consumption and reversing the
order of the digits. The ML algorithm Random Forest classifier gave the best result
over KNN, DT, Bagging and ANN for such theft detection [31]. Logistic Regression
and SVM performed the best among all supervised learning methods for analysing
audio signals and classifying them as drone or non-drone sounds for detection of
drones or other unmanned aerial vehicles (UAVs) using DREGON (Drone Egonoise
and Localization) datasets [32]. The use of a fleet of drones to connect a large number
of dispersed IoT devices is a novel scenario for the Industry 4.0 framework. The drone
fleet, taking into account fluctuating bandwidth demand and the availability of green
power. RL controls drones to make optimal decisions by learning from the envi-
ronment and dynamically adjusting fleet usage to match varying demands and the
availability of renewable energy sources [33].

4 Conclusion

This research study provides a thorough examination of a machine learning algo-

rithm for solving various engineering problems in Thermal Engineering, Energy
Applications, and Electrical Engineering.
• GA for feature selection and KNN for prediction of surface tension of ionic fluid
show precise accuracy.
• The XGBoost model is more accurate than other ML models in wind and solar
energy applications.
• A Support Vector Machine with a different kernel that can capture nonlinear
data patterns in higher dimensions is used as a prediction model for all three
engineering domains and is compared to the ANN model for its performance.
A Short Review of Machine Learning Techniques for Thermal, Energy … 13

• C4.5 trees (a variant of decision tree) outperform Neural Network for prediction
model of domestic appliance level of electricity during pandemic.
• RL model base controller is used to fleet the drone to find the optimal renewable
energy sources.
• To reduce the model complexity and computation power for engineering appli-
cations simple machine learning algorithms can be applied over complex neural
network model.
Growing hardware and software applications assist machine learning models in
solving complex problems. There is still research in areas such as human mimicry,
AI application replacement over sensors, and smaller models that can achieve the
best output model and many more.

References

1. Amershi S et al (2019) Software engineering for machine learning: a case study. In: 2019 IEEE/
ACM 41st international conference on software engineering: software engineering in practice
(ICSE-SEIP), May, pp 291–300. IEEE
2. Sharma N, Sharma R, Jindal N (2021) Machine learning and deep learning applications-a
vision. Glob Trans Proc 2(1):24–28
3. Arulkumaran K et al (2017) Deep reinforcement learning: a brief survey. IEEE Signal Process
Mag 34(6):26–38
4. Turet JG, Costa APCS (2022) Hybrid methodology for analysis of structured and unstructured
data to support decision-making in public security. Data Knowl Eng 141:102056
5. Schneider P, Xhafa F (2022) Machine learning: ML for eHealth systems. In: Anomaly detection
and complex event processing over IoT data streams, pp 149–191. Academic
6. Parashar A et al (2023) Data preprocessing and feature selection techniques in gait recognition:
a comparative study of machine learning and deep learning approaches. Pattern Recognit Lett
7. Shu L, Lu F, Chen Y (2023) Robust forecasting with scaled independent component analysis.
Financ Res Lett 51:103399
8. Yu K, Lin S, Guo GD (2023) Quantum dimensionality reduction by linear discriminant analysis.
Physica A 614:128554
9. Liu S, Chen Z, Jiao F (2023) Detection of maize seed germination rate based on improved
locally linear embedding. Comput Electron Agric 204:107514
10. Huang J, Wang F, Qiao L, Yang X (2023) T-distributed stochastic neighbor embedding echo
state network with state matrix dimensionality reduction for time series prediction. Eng Appl
Artif Intell 122:106055
11. Goh WWB, Hui HWH, Wong L (2023) How missing value imputation is confounded with
batch effects and what you can do about it. Drug Discov Today 103661
12. Sadrara M, Khorrami MK (2023) Principal component analysis–multivariate adaptive regres-
sion splines (PCA-MARS) and back propagation-artificial neural network (BP-ANN) methods
for predicting the efficiency of oxidative desulfurization systems using ATR-FTIR spec-
troscopy. Spectrochim Acta Part A: Mol Biomol Spectrosc 122944
13. Khurana D, Koli A, Khatter K, Singh S (2023) Natural language processing: state of the art,
current trends and challenges. Multimedia Tools Appl 82(3):3713–3744
14. Yalcin K, Cicekli I, Ercan G (2022) An external plagiarism detection system based on part-of-
speech (POS) tag n-grams and word embedding. Expert Syst Appl 197:116677
15. Nanni L, Maguolo G, Paci M (2020) Data augmentation approaches for improving animal
audio classification. Eco Inform 57:101084
14 N. R. Swain et al.

16. Sharma V et al (2021) Video processing using deep learning techniques: a systematic literature
review. IEEE Access 9:139489–139507
17. Khanum M et al (2015) A survey on unsupervised machine learning algorithms for automation,
classification and maintenance. Int J Comput Appl 119(13)
18. Blum A, Mitchell T (1998) Combining labeled and unlabeled data with co-training. In: Proceed-
ings of the eleventh annual conference on Computational learning theory, July 1998, pp
92–100
19. Chen Z, Yu S, Hu B, Hu R (2023) Multi-band and wide-angle nonreciprocal thermal radiation.
Int J Heat Mass Transf 209:124149
20. Abourehab MA et al (2022) Efficiency development of surface tension for different ionic liquids
through novel model of machine learning technique: application of in-thermal engineering. J
Mol Liq 367:120391
21. Cañada-Soriano M et al (2023) Application of machine learning algorithms in thermal images
for an automatic classification of lumbar sympathetic blocks. J Therm Biol 113:103523
22. Xing C et al (2023) Machine learning-based multi-objective optimization and thermody-
namic evaluation of organic Rankine cycle (ORC) system for vehicle engine under road
condition. Appl Therm Eng 120904
23. Boujelbene M et al (2023) Machine-learning optimization of an innovative design of a Li-ion
battery arrangement cooling system. J Energy Storage 58:106331
24. Song H, Song M, Liu X (2022) Online autonomous calibration of digital twins using machine
learning with application to nuclear power plants. Appl Energy 326:119995
25. Li Y, Yi YK (2023) Optimal shape design using machine learning for wind energy and pressure.
J Build Eng 70:106337
26. Lydia M, Kumar GEP (2021) Machine learning applications in wind turbine generating systems.
Mater Today: Proc 45:6411–6414
27. Said Z et al (2022) Exploring the specific heat capacity of water-based hybrid nanofluids for
solar energy applications: a comparative evaluation of modern ensemble machine learning
techniques. J Energy Storage 54:105230
28. Salari A et al (2023) A machine learning approach to optimize the performance of a combined
solar chimney-photovoltaic thermal power plant. Renew Energy 212:717–737
29. Condemi C et al (2021) Hydro-power production capacity prediction based on machine learning
regression techniques. Knowl-Based Syst 222:107012
30. Lentzas A, Vrakas D (2022) Machine learning approaches for non-intrusive home absence
detection based on appliance electrical use. Expert Syst Appl 210:118454
31. Zidi S, Mihoub A et al (2023) Theft detection dataset for benchmarking and machine learning
based classification in a smart grid environment. J King Saud Univ-Comput Inf Sci 35(1):13–25
32. Lacava G et al (2022) Drone audio recognition based on machine learning techniques. Procedia
Comput Sci 207:848–857
33. Faraci G, Raciti A et al (2020) Green wireless power transfer system for a drone fleet managed
by reinforcement learning in smart industry. Appl Energy 259:114204
Design of Intelligent ICT Irrigation
System Using Crop Growth Big Data
Analysis

Ssang-Hee Seo and Bong-Hyun Kim

Abstract The existing irrigation system was a method of utilizing the irrigation
system through input values based on user input. Therefore, it has been raised that
the automation system of the existing system is difficult to be introduced into a farm
environment with low technical capacity due to the difficulty of digitization of equip-
ment and periodic input. Therefore, in this paper, a customized, intelligent irrigation
system algorithm was designed using big data analysis based on the growth of culti-
vated crops. In addition, an irrigation system was designed according to temperature,
humidity, piping, light quantity, and water content of crops. Through this, the control
monitoring based on the recurrent neural network (RNN) was applied by utilizing
big data analysis. In this paper, we designed a solution that is easy to manage and
easy to use for cultivation and growth by using a customized, intelligent watering
system and various ICT sensors. In addition, through post-management, the system
was designed to provide easy usability to users with low technological acceptance by
changing S/W and major control devices according to changes in cultivated crops.
Designed and proposed in the paper, the customized and intelligent watering system
can be used to provide uncomplicated usage using initial modeling for crops. In
addition, due to the user-customized intelligent irrigation system, it is possible to
maintain the facility through simple monitoring through a system that continuously
self-feedback and decision-making even with a simple setup at the initial step of
installation.

Keywords Irrigation system · Big Data · Intelligent ICT · Crop growth

management · Monitoring

S.-H. Seo
Department of Computer Science and Engineering, Kyungnam University, 7 Kyungnamdaehak-ro,
Masanhappo-gu, Changwon-si, Gyeongsangnam-do 51767, Republic of Korea
e-mail: [email protected]
B.-H. Kim (B)
Department of Computer Engineering, Seowon University, 377-3 Musimseo-ro, Seowon-gu,
Cheongju-si, Chungcheongbuk-do 28674, Republic of Korea
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 15
A. S. Shastri et al. (eds.), Machine Learning and Optimization for Engineering Design,
Engineering Optimization: Methods and Applications,
https://doi.org/10.1007/978-981-99-7456-6_2
16 S.-H. Seo and B.-H. Kim

1 Introduction

Irrigation system means a system that supplies water to crops. In other words, agri-
cultural watering is to supply water so that proper moisture is maintained for normal
growth of crops. The reason why the irrigation system is important in the agricul-
tural environment is that it plays a role in properly supplying the life-like moisture
of the crop. Moisture supplied to the crop serves as a solvent for various inorganic
nutrients, and enables absorption and migration of nutrients. In addition, it enables
enzymes and metabolic functions, and is a basic material for photosynthesis [1]. The
moisture of the crops plays a role in keeping the body intact by squeezing the cells
and maintaining the body temperature through augmentation.
As such, it plays a very important role in properly supplying water in the process
of cultivating crops. Accordingly, irrigation systems for supplying water to crops
using various methods and devices have been developed and utilized. However, the
existing irrigation system is a level that utilizes the irrigation system through input
values based on user input. It has been raised that it is difficult to introduce existing
irrigation automation equipment into farming environments with low technological
capacity due to the difficulty of digitization of equipment and periodic input. In
addition, the existing automated irrigation system is not applied with intelligent
technology, and the user suffers a lot of inconvenience. That is, due to the difficulty
of agricultural workers due to the direct manipulation of the user and the absence of
a control system based on growth information, there is a hassle of irrigation by the
farm manpower [2–4].
After all, it is necessary to check the growth environment of the crop anytime,
anywhere based on accurate data on the growth information and environmental infor-
mation of the crop. In addition, it is necessary to implement an agricultural environ-
ment that can improve the productivity and quality of agricultural products even if
less labor, energy, and nutrients are introduced than before. To implement this, smart
farm solutions are continuously developed and introduced. In general, IoT, big data,
AI, automation systems, and robot technologies are applied to facility horticulture
(vinyl and glass greenhouses), and orchards to implement intelligent farm forms
that can remotely or automatically perform the maintenance and management of the
growing environment of crops. Techniques for doing this are being developed [5, 6].
Through the design of an intelligent irrigation system, it is possible to develop irri-
gation solutions that anyone can easily use through tailoring for agricultural workers
who are reluctant to introduce technology due to low technology acceptance or incon-
sistency with existing methods. In particular, it is possible to continuously implement
self-decision monitoring by developing an analysis and decision model of big data
acquired through RNN-based control monitoring S/W. In addition, it is possible to
derive the result values of cultivated crops such as harvest time and yield with the
analyzed data, and increase production by providing various information services to
users with recorded logs and learned data.
Therefore, in this paper, a customized, intelligent irrigation system algorithm was
designed using big data analysis based on the growth of cultivated crops. In addition,
Design of Intelligent ICT Irrigation System Using Crop Growth Big … 17

an irrigation system was designed according to temperature, humidity, piping, light

quantity, and water content of crops. Through this, the control monitoring based on
the recurrent neural network (RNN) was applied by utilizing big data analysis. In
addition, we designed a solution that is easy to manage and easy to use for cultivation
and growth by using a customized, intelligent watering system and various ICT
sensors. In addition, through post-management, the system was designed to provide
easy usability to users with low technological acceptance by changing S/W and major
control devices according to changes in cultivated crops. In this thesis, in Sect. 1,
the research outline, necessity, and purpose were written. In Sect. 2, the research
methodology was prepared through the current status and scale of ICT agricultural
technology. In Sect. 3, the design of the irrigation system proposed in the paper was
carried out, and finally, in Sect. 4, the research results and discussion were conducted.

2 ICT Agriculture Technology Trends

ICT agriculture solution checks the growth environment of crops anytime, anywhere
based on accurate data on crop growth information and environmental information,
and makes timely prescriptions to increase productivity and quality of agricultural
products even with less labor and energy nutrients than before a possible agricul-
tural technology integration solution. In other words, it is an integrated agricultural
technology solution that can appropriately maintain and manage the growing envi-
ronment of crops remotely or automatically by grafting ICT to a vinyl house or
orchard. This, convergence of agriculture and ICT can be applied in various forms
to promote innovation in the field from distribution to consumption and rural life in
addition to the production field. Domestic ICT agricultural technology solutions are
spreading to fields such as distribution and consumption, but until now, agricultural
production has been developed as the core. It is thought that it is concentrated in the
monitoring and control steps during production, and the development of optimization
algorithms using big data, automation technologies linked to robots, etc. are believed
to remain in the R&D step. Currently, ICT agricultural technology solutions applied
to domestic farms remain at the level of opening and controlling cultivation facilities
through smart media mainly based on environmental information [7–9].
In the case of the Netherlands, as a representative ICT agricultural technology
solution country, the land area is only half of that of Korea, but through the intro-
duction of ICT utilization, it is developing into the world’s second-largest agricul-
tural product exporting country. The Netherlands is a typical horticultural country,
where 99% of all greenhouses are glass greenhouses. Based on decades of accumu-
lated data and know-how in optimizing the cultivation environment, various sensors
and control solutions are being developed and utilized. The agricultural ICT tech-
nology aims to optimize production volume and quality. In addition, Dutch repre-
sentative company Priva produces the world’s best greenhouse environment control
system and exports it to countries around the world. In the U.S., it is attempting to
18 S.-H. Seo and B.-H. Kim

fully integrate nanotechnology, robotics, etc., into IoT, in agriculture. IoT, nanotech-
nology, and robot technology are converging to be used in agriculture. In the case
of Google, it is currently in the process of developing artificial intelligence decision
support system technology to collect seeds, fertilizers, and pesticides by collecting
big data on soil, moisture, and crop health. In Japan, companies such as IBM, NEC,
Fujitsu, and NTT provide various services by integrating ICT technology into the
agricultural field. Typical examples are IBM’s agricultural product history tracking
service, NEC’s M2M-based growth environment monitoring and logistics service,
and Fujitsu’s agricultural management cloud service system. Israel is developing as
a leader in the field of cultivation monitoring among many areas of ICT agricultural
technology solutions. Precise yield is predicted by automatically measuring crop
growth information such as crop size, stem change, and leaf temperature, and auto-
matically adjusting the watering cycle and water supply. In particular, the production
of crop stress sensors has increased production by more than 40% [10].
In recent years, ICT agricultural technology solutions that improve productivity
by combining ICT and BT with agriculture have attracted attention to solve the food
shortage caused by global warming. Advanced industrial countries such as Europe,
the United States, and Japan have applied ICT technology to agriculture to produce
crops accurately. The global market for ICT agricultural technology solutions is
estimated at $2.8 billion by 2015. It is also expected to grow to 11.8% per year
by 2020 and to $4.92 billion. In Korea, cases of introduction are increasing due to
government support and research and development by ICT companies. The domestic
ICT agricultural technology solution production market is growing at an average
annual rate of 10.5% from 2,429.5 billion won in 2012, and is expected to grow to
440 billion won by 2020. In 2020, the domestic ICT agricultural technology solution
production system market is estimated at 2.2 trillion won. This accounts for 41.6%
of the total market [11]. The intelligent agricultural machinery market is expected
to grow to 279.97 billion won (51.8%), and the plant-related market, which is the
initial step of market formation, will grow to 357.6 billion won (6.6%).
According to the domestic and foreign smart agriculture industry trend analysis
report, the technology level of ICT agricultural technology solutions is in the order
of Netherlands (99.1%), Japan (97.5%), Germany (93.3%), UK (89.5%), Australia
(83.4%), etc. The level is different. The level of technology in the United States, the
Netherlands and Japan is less than 0.5 years, and the gap is small compared to other
countries. At present, the level of ICT agricultural technology solutions in Korea
is estimated to be about 5% if the period is set to about 70% of the countries with
the best technology [12, 13]. Table 1 shows the level and gap of ICT agricultural
technology from around the world as of 2020, according to the domestic and foreign
smart agriculture industry trend analysis report. In Table 1, skill levels are expressed
in %, and skill gaps are expressed in years.
Design of Intelligent ICT Irrigation System Using Crop Growth Big … 19

Table 1 World ICT agricultural technology solution technology level and gab
Division Korea USA Japan England Netherland Germany Australia China
Tech. level 75.0 100 97.5 89.5 99.1 93.3 83.4 61.0
Tech. gap −5.2 0 −0.5 −2.3 0 −1.2 −3.6 −7.2

3 Irrigation System Design

The existing irrigation system utilizes an irrigation system through input values based
on user input. Such existing automation equipment has the hassle of digitizing and
periodic input of equipment. In addition, there is a disadvantage that it is difficult to
introduce it into a farm environment with low technical capacity. Therefore, in this
paper, we designed and developed an intelligent, customized irrigation algorithm
based on data. In addition, an irrigation system was designed according to tempera-
ture, humidity, piping, light quantity, and water content of crops. Finally, an integrated
irrigation system was designed to facilitate growth management and device control
by applying control monitoring based on RNN (recurrent neural network) using big
data.
Big data related to the growth of cultivated crops is largely collected and built
from two service platforms. First, growth big data is built using data collected
based on IoT sensors built in existing smart farms. Smart farm IoT sensors refer
to data collected through web cameras for environmental monitoring, indoor envi-
ronment sensors, weather observation sensors, energy control sensors, and irriga-
tion/fertilization control sensors. Second, data collected based on the monitoring
system for real-time response service means data collected from real-time cultiva-
tion information monitoring and smart farm operation management system. Figure 1
shows a schematic diagram of an integrated irrigation solution based on big data for
controlling the growth environment of cultivated crops.

Fig. 1 System design schematic

20 S.-H. Seo and B.-H. Kim

In order to design an intelligent irrigation system, the need for irrigation

customized for farms and an aging agricultural environment with low technological
acceptance should be applied, not an irrigation control system for various facility
cultivation. In addition, rather than interface of various technologies, it is necessary
to link with easy and universal technologies for a centralized system. That is, the ease
of installation of the facility, the continuity of real-time remote control, etc. must be
applied, the usability and convenience are improved, and the design of an intelli-
gent automated irrigation system in a decision-making method capable of learning
is needed [14]. To this end, in this paper, an integrated environmental control ICT
measurement sensor that can measure temperature, humidity, irrigation, light quan-
tity, and water content of a crop is applied to the system. In addition, the house
module was designed to enable customized facility cultivation, and it was designed
to be compatible with existing irrigation systems[15]. Lastly, a learning-type model
capable of decision-making through analysis of collected data was designed and
applied.
First, an environmental control ICT measurement integrated sensor capable of
complex measurements of temperature, humidity, piping, light quantity, and water
content of crops was designed in the system. The integrated ICT measurement sensor
performs environmental monitoring by measuring outdoor temperature, humidity
status, and weather information. In addition, a customized ICT module for growth and
cultivation for controlling automatic opening and closing devices such as irrigation,
side window, skylight, and sunshade membrane was interlocked according to the
measured data. The integrated ICT measurement sensor was designed in conjunction
with a data acquisition sensor for dangerous situations such as crop disease, abnormal
signs, intrusion, and fire. Figure 2 shows a schematic of data collection and control
in an intelligent, custom ICT irrigation system.
Second, the house module was designed to enable customized facility cultivation,
and it was designed to be compatible with existing irrigation systems. Through the
design of customized cultivated crops, the house module was applied so that crops
with different growth periods can be cultivated according to each period. In addition,
it was designed by applying attached facilities to enable interworking with existing
systems using facility equipment and modules that are easy to use and compatible.
Lastly, a learning-type model capable of decision-making through analysis of
collected data was designed and applied. Integrated SW for modeling and analyzing
acquired growth information was applied. That is, an optimal growth environment
is created by analyzing and learning through a computer. Through this, the SW
model was designed to enable decision-making. Also, through the development of a
recurrent neural network (RNN)-based learnable algorithm, a customized, intelligent
irrigation system SW that can be continuously trained was applied. Finally, we created
our own environment through computer-acquired data and designed a model that can
make decisions. In addition, the process was designed to enable real-time response
due to continuous log tracking of the decision model. Figure 3 shows the algorithm
and process for each step of learning.
The learning process of the intelligent and customized ICT irrigation system is
performed in the RNN model learning phase, the feedback detection phase, and
Design of Intelligent ICT Irrigation System Using Crop Growth Big … 21

Fig. 2 Intelligent irrigation system data collection and control

Fig. 3 The algorithm and process for each step of learning

22 S.-H. Seo and B.-H. Kim

the evolution phase. In addition, it reflects the learning result derived from the final
evolutionary step and re-enters the feedback detection step through linkage with real-
time monitoring data to control the irrigation system according to the environment.
Finally, we analyzed the acquired data, designed software for automatic control
with RNN-based computers capable of learning, analyzed acquired big data, and
designed a decision-making model. First, in the SW design for automatic control
with RNN-based computers capable of analyzing and learning acquired data, an
integrated SW that models and analyzes acquired growth information was designed.
By analyzing and learning through computer, the optimal growth environment was
created and the software capable of decision-making was designed. In addition,
it is proposed to design a computer SW that can learn continuously through the
development of a learnable algorithm based on a recurrent neural network (RNN).
Next, analysis of the acquired big data and design of a decision-making model were
performed. We designed a model that can make decisions by creating an environment
on its own through computer-generated data. In addition, we designed a real-time
coping process due to continuous log tracking of the decision-making model. In such a
model design, it is necessary to secure reliability through performance evaluation. To
this end, verification through sensor data collection and measurement cycle testing,
data integrity verification through data input and output repetition, control error
verification using a precision injection quantity measuring machine while measuring
injection quantity through sensors, and considering the stability of control devices
Performance evaluation such as verification should be linked.

4 Conclusion

In the smart farm solution, the design of the irrigation system is done through
customized design for each growing season, so it is very important for the envi-
ronment creation of the cultivated crop. That is, it is necessary to design and develop
an intelligent irrigation control system capable of designing in accordance with the
environment of facility cultivation in a control system method according to the growth
environment. However, most farms using domestic smart farms are operating smart
farms using excellent products from foreign countries such as the Netherlands. As
foreign products are used, DBs such as growth information accumulated in smart
farms are naturally being supplied to foreign products. In addition, if the growth
information DB accumulated in the domestic smart farm supplied to foreign prod-
ucts is leaked, a huge blow to domestic farming occurs. As a result, the current status
of domestic farming has been leaked overseas, affecting the production and supply
of domestic agricultural products, making domestic farming more difficult. There-
fore, it is necessary to design and develop an intelligent smart farm irrigation system
as it is necessary to create a smart farm with domestic products to facilitate post-
management such as A/S and to prevent DB leakage of the growth environment.
However, the existing automated irrigation system does not apply the intelligent
method, which causes difficulties for agricultural workers due to direct manipulation
Design of Intelligent ICT Irrigation System Using Crop Growth Big … 23

by the user. In addition, the lack of a control system based on the growth information
has been raised as a problem because the farmer must adjust the irrigation directly.
Therefore, in this paper, a customized, intelligent irrigation system algorithm was
designed using big data analysis based on the growth of cultivated crops. In addition,
an irrigation system was designed according to temperature, humidity, piping, light
quantity, and water content of crops. Through this, the control monitoring based on
the recurrent neural network (RNN) was applied by utilizing big data analysis. In
addition, we designed a solution that is easy to manage and easy to use for cultivation
and growth by using a customized, intelligent watering system and various ICT
sensors. In addition, through post-management, the system was designed to provide
easy usability to users with low technological acceptance by changing S/W and
major control devices according to changes in cultivated crops. To this end, in this
paper, an integrated environmental control ICT measurement sensor that can measure
temperature, humidity, irrigation, light quantity, and water content of a crop is applied
to the system. In addition, the house module was designed to enable customized
facility cultivation, and it was designed to be compatible with existing irrigation
systems. Finally, a learning-type model capable of decision-making was designed
and applied through the analysis of data collected through the designed irrigation
system.
In future research, it is necessary to develop and connect technologies to prevent
DB leakage, such as smart farm growth environments. If it is not self-developed,
DB such as growth information accumulated in smart farms can also be leaked to
supply product companies. Such leakage of growth information can cause enormous
problems to the agricultural system itself. In addition, the agricultural status may
be leaked, which may adversely affect the production and supply of agricultural
products. In the end, it is necessary to prevent DB leakage such as smart farm growth
environment, and research on smart farm irrigation/management systems that have
established an information security system for future big data utilization is necessary.

References

1. Yang A, Kim J-G (2018) An architecture and design of data converter for IoT-based smart
farm. Int J Smart Home 12(4):7–12
2. Zhang Y, Xiang Y, Wang L (2017) Power system reliability assessment incorporating cyber
attacks against wind farm energy management systems. IEEE Trans Smart Grid 8(5):2343–
2357
3. Choudhari P, Borse A, Chauhan H (2018) Smart irrigation and remote farm monitoring system.
Int J Comput Appl 180(38):24–26
4. Phanthuna N, Lumnium T (2017) Design and application for a smart farm in Thailand based
on IoT. Appl Mech Mater 866:433–438
5. Lee S-G, Cho B-H (2018) Sign of scalable sensor and actuator interface module for smart farm.
Int J Smart Home 12(4):1–6
6. Balakrishnan M, Arul Antony S, Gunasekaran S, Natarajan RK (2008) Impact of dyeing
industrial effluents on the groundwater quality in Kancheepuram. Indian J Sci Techno 1(7):1–8
7. Kesavan KG, Parameswari R (2005) Evaluation of groundwater quality in Kancheepuram.
Indian J Environ Prot 25(3):235–239
24 S.-H. Seo and B.-H. Kim

8. Noguchi N (2016) Remote sensing technology for ICT agriculture. J Robot Soc Jpn 34(2):100–
102
9. Swaminathan M, Swaminathan MS (2018) ICT and agriculture. CSI Trans ICT 6(3–4):227–229
10. Voogt J, Pelgrum H (2005) ICT and curriculum change. Hum Technol: Interdiscip J Hum ICT
Environ 1(2):157–175
11. Singh S, Ahlawatat S, Sanwal S (2017) Role of ICT in agriculture: policy implications. Orient
J Comput Sci Technol 10(3):691–697
12. Rohila AK, Yadav K, Ghanghas BS (2017) Role of information and communication technology
(ICT) in agriculture and extension. J Appl Nat Sci 9(2):1097–1100
13. Clemmens AJ (1991) Irrigation uniformity relationships for irrigation system management. J
Irrig Drain Eng 117(5):682–699
14. Pandya AB (2019) Solar powered irrigation systems. Irrig Drain 68(2):379–380
15. Han C, Zhang B, Liu Y (2020) Efficient water-saving irrigation based on regional irrigation
schedule optimization. Desalin Water Treat 187:30–41
OpenCV and MQTT Based Intelligent
Management System

Anand Mahajan, Satej Gadekar, Sumit Sagave, Smita Paithankar,

Bhushan Rakhonde, and Chetan Khadse

Abstract In this paper, a system is proposed which is intelligent and can perform
identification, counting, and calculation of density of vehicles. After calculating the
traffic density, the system classifies the density into low, medium, and high density
with the help of a decision algorithm. This system is based on Python programming,
and the libraries used in Python are Open-source Computer Vision, NumPy, Chardet,
and time library. The system is implemented in IoT-based platform Message Queuing
Telemetry Transport. The system methodology is carried out in four phases. The first
is vehicle detection and counting. In the second phase, the number plate of the vehicle
is detected as well as displayed. The third phase includes traffic density detection
and finding out the emergency vehicle based on GPS Tracking using ESP32 and IoT
over MQTT. Last phase is related to pollution monitoring. This complete model for
the system is made, and the results are shown in the paper.

Keywords Chardet · NumPy · OpenCV python · Pip · Time · Wi-Fi · Pub-sub

client · Software serial · TinyGPS++

1 Introduction

The smart traffic management system is a centralized system that works collectively
with sensors and makes traffic congestion easy for everyday travelers. Many countries
have poor traffic handling and management. The conventional traffic signal needs to
be upgraded with the latest technology to mitigate the increasing number of vehicles

A. Mahajan (B)
Delft University of Technology, Delft, The Netherlands
e-mail: [email protected]
S. Gadekar · S. Sagave · S. Paithankar · C. Khadse
Dr. Vishwanath Karad MIT World Peace University Pune, Pune, Maharashtra, India
B. Rakhonde
Shri Sant Gajanan Maharaj College of Engineering Shegaon, Shegaon, Maharashtra, India

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 25
A. S. Shastri et al. (eds.), Machine Learning and Optimization for Engineering Design,
Engineering Optimization: Methods and Applications,
https://doi.org/10.1007/978-981-99-7456-6_3
26 A. Mahajan et al.

on roads. The installation of smart traffic management systems has created a unique
way to detect speed violating vehicles. However, many of the existing methods of
vehicle detection have a scope for improvement. Camera input is required to teach
the system to identify the car. The automatic number plate detection system uses
Image subtraction, character segmentation, and further analysis as a founding base
for vehicle tracking [1–3]. OpenCV framework is used to capture and process each
frame per second and is accordingly used for further analysis. While detecting and
counting the number of cars, a threshold line must be set as a reference line for
the system. This line is drawn at an optimum level to count the vehicles only after
crossing the line [4–9].
Vehicles are the most common things causing the ever-increasing pollution on
this planet. Therefore, it is crucial to track the increase in pollution daily and use the
data to reduce pollution in every possible way. Hence, the last phase in this paper is
to collect the pollution data using the MQ135 gas sensor, one of the most common
sensors used to detect gases such as CO2, Smoke, NH3, NOx, Alcohol, Benzene,
etc. [10, 11].
In this paper, two methods have been discussed to detect the arrival of emergency
vehicles.
In the first method, as soon as the vehicle comes near the traffic signal, the driver
can indicate its arrival using an android application by connecting with the Bluetooth
receiver situated in the traffic signal. Then, an android application is used to change
the traffic signal. Thus, a simple android application is used to indicate the arrival
of the emergency vehicle. However, due to the limited range of Bluetooth zones, a
more feasible method is to keep track of the emergency vehicle.
The second method requires low-power sensors, a microcontroller, and a robust
IoT platform. In this case, a NodeMCU ESP32, a popular Wi-Fi-based microcon-
troller, is used. The vehicle’s location is continuously tracked using the GPS module
[12, 13]. The real-time location is sent to the MQTT server, which further sends
the information of the arriving vehicle to the receiver at traffic signals. The operator
can then remotely change the traffic signal and allow easy access to the emergency
vehicle [9, 14].

2 Methodology

Figure 1 represents the block diagram for the working of the proposed in a smart
traffic management system. Firstly, the high-resolution cameras fitted on the traffic
signal will detect and capture the number plate of vehicles commuting on the road-
ways. Secondly, the proposed system detects and counts the number of incoming and
outgoing vehicles using OpenCV framework, which is further used to determine the
density of vehicles in each lane.
Furthermore, based on the collected data from the previous steps and the GPS
location of the emergency vehicle a decision is made to control the traffic signal
remotely via android application over MQTT without the need for the emergency
Exploring the Variety of Random
Documents with Different Content
Erckmann-Chatrian, Brothers Rantzau.
Fenn (G. Manville) Off to the Wilds.
—— Silver Cañon.
Groves (Percy) Charmouth Grange; a Tale of the
17th Century.
Heldmann (B.) Mutiny on Board the Ship Leander.
Henty (G. A.) Cornet of Horse: a Tale of
Marlborough's Wars.
—— Jack Archer; a Tale of the Crimea.
—— Winning his Spurs: a Tale of the Crusades.
Johnstone (D. Lawson) Mountain Kingdom.
Kennedy (E. B.) Blacks and Bushrangers in
Queensland.
Kingston (W. H. G.) Ben Burton; or, Born and Bred
at Sea.
—— Captain Mugford; or, Our Salt and Fresh
Water Tutors.
—— Dick Cheveley.
—— Heir of Kilfinnan.

Kingston (W. H. G.) Snowshoes and Canoes.

—— Two Supercargoes.
—— With Axe and Rifle on the Western Prairies.
Laurie (A.) Conquest of the Moon.
—— New York to Brest in Seven Hours.
MacGregor (John) A Thousand Miles in the Rob
Roy Canoe on Rivers and Lakes of Europe.
Maclean (H. E.) Maid of the Ship Golden Age.
Meunier, Great Hunting Grounds of the World.
Müller, Noble Words and Deeds.
Perelaer, The Three Deserters; or, Ran Away from
the Dutch.
Reed (Talbot Baines) Sir Ludar: a Tale of the Days
of the Good Queen Bess.
Rousselet (Louis) Drummer-boy: a Story of the
Time of Washington.
—— King of the Tigers.
—— Serpent Charmer.
—— Son of the Constable of France.
Russell (W. Clark) Frozen Pirates.
Stanley, My Kalulu—Prince, King and Slave.
Winder (F. H.) Lost in Africa.
Low's Standard Series of Books by popular writers,
cloth gilt, 2s.; gilt edges, 2s. 6d. each.
Alcott (L. M.) A Rose in Bloom.
—— An Old-Fashioned Girl.
—— Aunt Jo's Scrap Bag.
—— Eight Cousins, illust.
—— Jack and Jill.
—— Jimmy's Cruise.
—— Little Men.
—— Little Women and Little Women Wedded.
—— Lulu's Library, illust.
—— Shawl Straps.
—— Silver Pitchers.
—— Spinning-Wheel Stories.
—— Under the Lilacs, illust.
—— Work and Beginning Again, ill.
Alden (W. L.) Jimmy Brown, illust.
—— Trying to Find Europe.
Bunyan (John) Pilgrim's Progress, (extra volume),
gilt, 2s.
De Witt (Madame) An Only Sister.
Francis (Francis) Eric and Ethel, illust.
Holm (Saxe) Draxy Miller's Dowry.
Jerdon (Gert.) Keyhole Country, illust.
Robinson (Phil) In My Indian Garden.
—— Under the Punkah.
Roe (E. P.) Nature's Serial Story.
Saintine, Picciola.
Samuels, Forecastle to Cabin, illust.
Sandeau (Jules) Seagull Rock.
Stowe (Mrs.) Dred.
—— Ghost in the Mill, &c.
—— My Wife and I.
—— We and our Neighbours.
See also Low's Standard Series.
Tooley (Mrs.) Life of Harriet Beecher Stowe.
Warner (C. Dudley) In the Wilderness.
—— My Summer in a Garden.
Whitney (Mrs.) A Summer in Leslie Goldthwaite's
Life.
—— Faith Gartney's Girlhood.
—— Hitherto.
—— Real Folks.
—— The Gayworthys.
—— We Girls.
—— The Other Girls: a Sequel.
*** A new illustrated list of books for boys and
girls, with portraits of celebrated authors, sent
post free on application.
LOWELL, J. R., Among my Books, Series I. and II.,
7s. 6d. each.
—— My Study Windows, n. ed. 1s.
—— Vision of Sir Launfal, illus. 63s.
MACDONALD, A., Our Sceptred Isle, 3s. 6d.
—— D., Oceania, 6s.

MACDONALD, Geo., Castle Warlock, a Homely

Romance, 3 vols. 31s. 6d.
—— See also Low's Standard Novels.
—— Sir John A., Life.
MACDOWALL, Alex. B., Curve Pictures of London,
1s.
MACGAHAN, J. A., Oxus, 7s. 6d.
MACGOUN, Commercial Correspondence, 5s.
MACGREGOR, J., Rob Roy in the Baltic, n. ed. 3s.
6d. and 2s. 6d.
—— Rob Roy Canoe, new edit., 3s. 6d. and 2s. 6d.
—— Yawl Rob Roy, new edit., 3s. 6d. and 2s. 6d.
MACKENNA, Brave Men in Action, 10s. 6d.
MACKENZIE, Sir Morell, Fatal Illness of Frederick
the Noble, 2s. 6d.
MACKINNON and SHADBOLT, South African
Campaign, 50s.
MACLAREN, A. See Preachers.
MACLEAN, H. E. See Low's Standard Books.
MACMASTER. See Low's Standard Novels.
MACMURDO, E., History of Portugal, 21s.; II. 21s.;
III. 21s.
MAHAN, A. T., Influence of Sea Power on History,
18s.
Maid of Florence, 10s. 6d.
MAIN, Mrs., High Life, 10s. 6d.
—— See also Burnaby, Mrs.
MALAN, A. N., Cobbler of Cornikeranium, 5s.
—— C. F. DE M., Eric and Connie's Cruise, 5s.
Man's Thoughts. See Gentle Life Series.
MANLEY, J. J., Fish and Fishing, 6s.

MANTEGNA and FRANCIA. See Great Artists.

MARCH, F. A., Comparative Anglo-Saxon Grammar,
12s.
—— Anglo-Saxon Reader, 7s. 6d.
MARKHAM, Adm., Naval Career, 14s.
—— Whaling Cruise, new edit. 7s. 6d.
—— C. R., Peru. See Foreign Countries.
—— Fighting Veres, 18s.
—— War Between Peru and Chili, 10s. 6d.
MARSH, G. P., Lectures on the English Language,
18s.
—— Origin and History of the English Language,
18s.
MARSHALL, W. G., Through America, new edit. 7s.
6d.
MARSTON, E., How Stanley wrote "In Darkest
Africa," 1s.
—— See also Amateur Angler, Frank's Ranche, and
Fresh Woods.
—— W., Eminent Actors, n. ed. 6s.
MARTIN, J. W., Float Fishing and Spinning, new
edit. 2s.
Massage. See Nursing Record Series.
MATTHEWS, J. W., Incwadi Yami, 14s.
MAURY, M. F., Life, 12s. 6d.
—— Physical Geography and Meteorology of the
Sea, new ed. 6s.
MEISSNER, A. L., Children's Own German Book
(Low's Series), 1s. 6d.
—— First German Reader (Low's Series), 1s. 6d.
—— Second German Reader (Low's Series), 1s.
6d.
MEISSONIER. See Great Artists.
MELBOURNE, Lord. See Prime Ministers.
MELIO, G. L., Swedish Drill, 1s. 6d.
MENDELSSOHN Family, 1729-1847, Letters and
Journals, 2 vols. 30s.; new edit. 30s.
—— See also Great Musicians.
MERRIFIELD, J., Nautical Astronomy, 7s. 6d.
MERRYLEES, J., Carlsbad, 7s. 6d. and 9s.
MESNEY, W., Tungking, 3s. 6d.
Metal Workers' Recipes and Processes, by W. T.
Brannt, 12s. 6d.
MEUNIER, V. See Low's Standard Books.
Michelangelo. See Great Artists.
MILFORD, P. Ned Stafford's Experiences, 5s.
MILL, James. See English Philosophers.
MILLS, J., Alternative Elementary Chemistry, 1s.
6d.
—— Chemistry Based on the Science and Art
Syllabus, 2s. 6d.
—— Elementary Chemistry, answers, 2 vols. 1s.
each.
MILTON'S Allegro. See Choice Editions.
MITCHELL, D. G. (Ik. Marvel) English Lands,
Letters and Kings, 2 vols. 6s. each.
—— Writings, new edit. per vol. 5s.
MITFORD, J., Letters, 3s. 6d.
—— Miss, Our Village, illust. 5s.
Modern Etchings, 63s. & 31s. 6d.
MOLLETT, J. W., Dictionary of Words in Art and
Archæology, illust. 15s.
—— Etched Examples, 31s. 6d. and 63s.
—— See also Great Artists.

MONCK. See English Philosophers.

MONEY, E., The Truth About America, 5s.; new
edit. 2s. 6d.
MONKHOUSE. See G. Artists.
Montaigne's Essays, revised by J. Hain Friswell, 2s.
6d.
—— See Gentle Life Series.
MOORE, J. M., New Zealand for Emigrant, Invalid,
and Tourist, 5s.
MORFILL, W. R., Russia, 3s. 6d.
MORLEY, Henry, English Literature in the Reign of
Victoria, 2s. 6d.
—— Five Centuries of English Literature, 2s.
MORSE, E. S., Japanese Homes, new edit. 10s. 6d.
MORTEN, Hospital Life, 1s.
MORTIMER, J., Chess Player's Pocket-Book, new
edit. 1s.
MORWOOD, V. S., Our Gipsies, 18s.
MOSS, F. J., Great South Sea, 8s. 6d.
MOSSMAN, S., Japan, 3s. 6d.
MOTTI, Pietro, Elementary Russian Grammar, 2s.
6d.
—— Russian Conversation Grammar, 5s.; Key, 2s.
MOULE, H. C. G., Sermons, 3s. 6d.
MOXLEY, West India Sanatorium, and Barbados,
3s. 6d.
MOXON, W., Pilocereus Senilis, 3s. 6d.
MOZART, 3s. Gr. Musicians.
MULLER, E. See Low's Standard Books.
MULLIN, J. P., Moulding and Pattern Making, 12s.
6d.
MULREADY, 3s. 6d. Great Artists.
MURILLO. See Great Artists.
MUSGRAVE, Mrs. See Low's Standard Novels.
—— Savage London, n. e. 3s. 6d.
My Comforter, &c., Religious Poems, 2s. 6d.
Napoleon I. See Bayard Series.
Napoleon I. and Marie Louise, 7s. 6d.
NELSON, Wolfred, Panama, 6s.
Nelson's Words and Deeds, 3s. 6d.
NETHERCOTE, Pytchley Hunt, 8s. 6d.
New Democracy, 1s.
New Zealand, chromos, by Barraud, 168s.
NICHOLSON, British Association Work and
Workers, 1s.
Nineteenth Century, a Monthly Review, 2s. 6d. per
No.
NISBET, Hume, Life and Nature Studies, 6s.
NIXON, Story of the Transvaal, 12s. 6d.
Nordenskiöld's Voyage, trans. 21s.
NORDHOFF, C., California, new edit. 12s. 6d.
NORRIS, Rachel, Nursing Notes, 2s.
NORTH, W., Roman Fever, 25s.
Northern Fairy Tales, 5s.
NORTON, C. L., Florida, 5s.
NORWAY, G., How Martin Drake Found his Father,
illus. 5s.
NUGENT'S French Dictionary, new edit. 3s.
Nuggets of the Gouph, 3s.
Nursing Record Series, text books and manuals.
Edited by Charles F. Rideal.
1. Lectures to Nurses on Antiseptics in Surgery. By
E. Stanmore Bishop. With coloured plates, 2s.
2. Nursing Notes. Medical and Surgical
information. For Hospital Nurses, &c. With
illustrations and a glossary of terms. By Rachel
Norris (née Williams), late Acting Superintendent
of Royal Victoria Military Hospital at Suez, 2s.
3. Practical Electro-Therapeutics. By Arthur
Harries, M.D., and H. Newman Lawrence. With
photographs and diagrams, 1s. 6d.
4. Massage for Beginners. Simple and easy
directions for learning and remembering the
different movements. By Lucy Fitch, 1s.
O'BRIEN, Fifty Year of Concession to Ireland, vol.
i. 16s.; vol. ii. 16s.
—— Irish Land Question, 2s.
OGDEN, James, Fly-tying, 2s. 6d.
O'GRADY, Bardic Literature of Ireland, 1s.
—— History of Ireland, vol. i. 7s. 6d.; ii. 7s. 6d.
Old Masters in Photo. 73s. 6d.
Orient Line Guide, new edit. 2s. 6d.
ORLEBAR, Sancta Christina, 5s.
Other People's Windows. See Gentle Life Series.
OTTÉ, Denmark and Iceland, 3s. 6d. Foreign
Countries.
Our Little Ones in Heaven, 5s.
Out of School at Eton, 2s. 6d.
OVERBECK. See Great Artists.
OWEN, Douglas, Marine Insurance, 15s.
Oxford Days, by a M.A., 2s. 6d.
PALGRAVE, Chairman's Handbook, new edit. 2s.
—— Oliver Cromwell, 10s. 6d.
PALLISER, China Collector's Companion, 5s.
—— History of Lace, n. ed. 21s.
PANTON, Homes of Taste, 2s. 6d.
PARKE, Emin Paska Relief Expedition, 21s.
PARKER, E. H., Chinese Account of the Opium
War, 1s. 6d.
PARSONS, J., Principles of Partnership, 31s. 6d.
—— T. P., Marine Insurance, 2 vols. 63s.
PEACH, Annals of Swainswick, 10s. 6d.
Peel. See Prime Ministers.
PELLESCHI, G., Gran Chaco, 8s. 6d.
PENNELL, H. C., Fishing Tackle, 2s.
—— Sporting Fish, 15s. & 30s.
Penny Postage Jubilee, 1s.
PERRY, Nora, Another Flock of Girls, illus. by Birch
& Copeland, 7s. 6d.
Peru, 3s. 6d. Foreign Countries.
PHELPS, E. S., Struggle for Immortality, 5s.
—— Samuel, Life, by W. M. Phelps and Forbes-
Robertson, 12s.
PHILLIMORE, C. M., Italian Literature, new. edit.
3s. 6d.
PHILLIPPS, W. M., English Elegies, 5s.
PHILLIPS, L. P., Dictionary of Biographical
Reference, new. edit. 25s.
—— W., Law of Insurance, 2 vols. 73s. 6d.
PHILPOT, H. J., Diabetes Mellitus, 5s.
—— Diet Tables, 1s. each.
Picture Gallery of British Art. I. to VI. 18s. each.
—— Modern Art, 3 vols. 31s. 6d. each.
PINTO, How I Crossed Africa, 2 vols. 42s.
Playtime Library. See Humphrey and Huntingdon.
Pleasant History of Reynard the Fox, trans. by T.
Roscoe, illus. 7s. 6d.
POCOCK, R., Gravesend Historian, 5s.
POE, by E. C. Stedman, 3s. 6d.
—— Raven, ill. by G. Doré, 63s.
Poems of the Inner Life, 5s.
Poetry of Nature. See Choice Editions.
Poetry of the Anti-Jacobin, 7s. 6d. and 21s.
POOLE, Somerset Customs and Legends, 5s.
—— S. Lane, Egypt, 3s. 6d. Foreign Countries.
POPE, Select Poetical Works, (Bernhard Tauchnitz
Collection), 2s.
PORCHER, A., Juvenile French Plays, 1s.
Portraits of Racehorses, 4 vols. 126s.
POSSELT, Structure of Fibres, 63s.
—— Textile Design, illust. 28s.
POYNTER. See Illustrated Text Books.
Preachers of the Age, 3s. 6d. ea.
Living Theology, by His Grace the Archbishop of
Canterbury.
The Conquering Christ, by Rev. A. Maclaren.
Verbum Crucis, by the Bishop of Derry.
Ethical Christianity, by H. P. Hughes.
Sermons, by Canon W. J. Knox-Little.
Light and Peace, by H. R. Reynolds.
Faith and Duty, by A. M. Fairbairn.
Plain Words on Great Themes, by J. O. Dykes.
Sermons, by the Bishop of Ripon.
Sermons, by Rev. C. H. Spurgeon.
Agoniæ Christi, by Dean Lefroy, of Norwich.
Sermons, by H. C. G. Moule, M.A.
Volumes will follow in quick succession by other
well-known men.
Prime Ministers, a series of political biographies,
edited by Stuart J. Reid, 3s. 6d. each.

1. Earl of Beaconsfield, by J. Anthony Froude.

2. Viscount Melbourne, by Henry Dunckley
("Verax").
3. Sir Robert Peel, by Justin McCarthy.
4. Viscount Palmerston, by the Marquis of
Lorne.
5. Earl Russell, by Stuart J. Reid.
6. Right Hon. W. E. Gladstone, by G. W. E.
Russell.
7. Earl of Aberdeen, by Sir Arthur Gordon.
8. Marquis of Salisbury, by H. D. Traill.
9. Earl of Derby, by George Saintsbury.
*** An edition, limited to 250 copies, is issued
on hand-made paper, medium 8vo, bound in
half vellum, cloth sides, gilt top. Price for the
9 vols. 4l. 4s. nett.

Prince Maskiloff. See Low's Standard Novels.

Prince of Nursery Playmates, new edit. 2s. 6d.
PRITT, T. N., Country Trout Flies, 10s. 6d.
Reynolds. See Great Artists.
Purcell. See Great Musicians.
QUILTER, H., Giotto, Life, &c. 15s.
RAMBAUD, History of Russia, new edit., 3 vols.
21s.
RAPHAEL. See Great Artists.
REDFORD, Sculpture. See Illustrated Text-books.
REDGRAVE, Engl. Painters, 10s. 6d. and 12s.
REED, Sir E. J., Modern Ships of War, 10s. 6d.
—— T. B., Roger Ingleton, Minor, 5s.
—— Sir Ludar. See Low's Standard Books.
REID, Mayne, Capt., Stories of Strange Adventures,
illust. 5s.
—— Stuart J. See Prime Ministers.
—— T. Wemyss, Land of the Bey, 10s. 6d.
Remarkable Bindings in British Museum, 168s.;
94s. 6d.; 73s. 6d. and 63s.
REMBRANDT. See Great Artists.
Reminiscences of a Boyhood, 6s.
REMUSAT, Memoirs, Vols. I. and II. new ed. 16s.
each.
—— Select Letters, 16s.
REYNOLDS. See Gr. Artists.
—— Henry R., Light & Peace, &c. Sermons, 3s. 6d.
RICHARDS, J. W., Aluminium, new edit. 21s.
RICHARDSON, Choice of Books, 3s. 6d.
RICHTER, J. P., Italian Art, 42s.
—— See also Great Artists.
RIDDELL. See Low's Standard Novels.
RIDEAL, Women of the Time, 14s.
RIFFAULT, Colours for Painting, 31s. 6d.
RIIS, How the Other Half Lives, 10s. 6d.
RIPON, Bp. of. See Preachers.
ROBERTS, Miss, France. See Foreign Countries.
—— W., English Bookselling, earlier history, 7s. 6d.
ROBIDA, A., Toilette, coloured, 7s. 6d.
ROBINSON, "Romeo" Coates, 7s. 6d.
—— Noah's Ark, n. ed. 3s. 6d.
—— Sinners & Saints, 10s. 6d.
—— See also Low's Standard Series.
—— Wealth and its Sources, 5s.
—— W. C., Law of Patents, 3 vols. 105s.
ROCHEFOUCAULD. See Bayard Series.
ROCKSTRO, History of Music, new ed. 14s.
RODRIGUES, Panama Canal, 5s.
ROE, E. P. See Low's Standard Series.
ROGERS, S. See Choice Editions.
ROLFE, Pompeii, 7s. 6d.
Romantic Stories of the Legal Profession, 7s. 6d.
ROMNEY. See Great Artists.
ROOSEVELT, Blanche R. Home Life of Longfellow,
7s. 6d.
ROSE, J., Mechanical Drawing, 16s.
—— Practical Machinist, new ed. 12s. 6d.
—— Key to Engines, 8s. 6d.
—— Modern Steam Engines, 31s. 6d.
—— Steam Boilers, 12s. 6d.
Rose Library. Popular Literature of all countries,
per vol. 1s., unless the price is given.
Alcott (L. M.) Eight Cousins, 2s.; cloth, 3s. 6d.
—— Jack and Jill, 2s.; cloth, 5s.
—— Jimmy's cruise in the Pinafore, 2s.; cloth, 3s.
6d.
—— Little Women.
—— Little Women Wedded; Nos. 4 and 5 in 1 vol.
cloth, 3s. 6d.
—— Little Men, 2s.; cloth gilt, 3s. 6d.
Alcott (L. M.) Old-fashioned Girls, 2s.; cloth, 3s.
6d.
—— Rose in Bloom, 2s.; cl. 3s. 6d.
—— Silver Pitchers.
—— Under the Lilacs, 2s.; cloth, 3s. 6d.
—— Work, A Story of Experience, 2 vols. in 1,
cloth, 3s. 6d.
Stowe (Mrs.) Pearl of Orr's Island.
—— Minister's Wooing.
—— We and Our Neighbours, 2s.
—— My Wife and I, 2s.
Dodge (Mrs.) Hans Brinker, or, The Silver Skates,
1s.; cloth, 5s.; 3s. 6d.; 2s. 6d.
Lowell (J. R.) My Study Windows.
Holmes (Oliver Wendell) Guardian Angel, cloth, 2s.
Warner (C. D.) My Summer in a Garden, cloth, 2s.
Stowe (Mrs.) Dred, 2s.; cloth gilt, 3s. 6d.
Carleton (W.) City Ballads, 2 vols. in 1, cloth gilt,
2s. 6d.
—— Legends, 2 vols. in 1, cloth gilt, 2s. 6d.
—— Farm Ballads, 6d. and 9d.; 3 vols. in 1, cloth
gilt, 3s. 6d.
—— Farm Festivals, 3 vols. in 1, cloth gilt, 3s. 6d.
—— Farm Legends, 3 vols. in 1, cloth gilt, 3s. 6d.
Clients of Dr. Bernagius, 2 vols.
Howells (W. D.) Undiscovered Country.
Clay (C. M.) Baby Rue.
—— Story of Helen Troy.
Whitney (Mrs.) Hitherto, 2 vols. cloth, 3s. 6d.
Fawcett (E.) Gentleman of Leisure.
Butler, Nothing to Wear.
ROSS, Mars, Cantabria, 21s.
ROSSINI, &c., See Great Musicians.
Rothschilds, by J. Reeves, 7s. 6d.
Roughing it after Gold, by Rux, new edit. 1s.
ROUSSELET. See Low's Standard Books.
ROWBOTHAM, F. J., Prairie Land, 5s.
Royal Naval Exhibition, a souvenir, illus. 1s.
RUBENS. See Great Artists.
RUGGLES, H. J., Shakespeare's Method, 7s. 6d.
RUSSELL, G. W. E., Gladstone. See Prime
Ministers.
—— W. Clark, Mrs. Dines' Jewels, 2s. 6d.
—— Nelson's Words and Deeds, 3s. 6d.
—— Sailor's Language, illus. 3s. 6d.
—— See also Low's Standard Novels and Sea
Stories.
—— W. Howard, Prince of Wales' Tour, illust. 52s.
6d. and 84s.
Russia. See Foreign Countries.
Saints and their Symbols, 3s. 6d.
SAINTSBURY, G., Earl of Derby. See Prime
Ministers.
SAINTINE, Picciola, 2s. 6d. and 2s. See Low's
Standard Series.
SALISBURY, Lord. See Prime Ministers.
SAMUELS. See Low's Standard Series.
SANDARS, German Primer, 1s.
SANDEAU, Seagull Rock, 2s. and 2s. 6d. Low's
Standard Series.
SANDLANDS, How to Develop Vocal Power, 1s.
SAUER, European Commerce, 5s.
—— Italian Grammar (Key, 2s.), 5s.
—— Spanish Dialogues, 2s. 6d.
—— Spanish Grammar (Key, 2s.), 5s.
—— Spanish Reader, new edit. 3s. 6d.
SAUNDERS, J., Jaspar Deane, 10s. 6d.
SCHAACK, M. J., Anarchy, 16s.
SCHAUERMANN, Ornament for technical schools,
10s. 6d.
SCHERER, Essays in English Literature, by G.
Saintsbury, 6s.
SCHERR, English Literature, history, 8s. 6d.
SCHIILLER'S Prosa, selections by Buchheim. Low's
Series, 2s. 6d.
SCHUBERT. See Great Musicians.
SCHUMANN. See Great Musicians.
SCHWEINFURTH. See Low's Standard Library.
Scientific Education of Dogs, 6s.
SCOTT, Leader, Renaissance of Art in Italy, 31s. 6d.
—— See also Illust. Text-books.
—— Sir Gilbert, Autobiography, 18s.
—— W. B. See Great Artists.
SELMA, Robert, Poems, 5s.
SERGEANT, L. See Foreign Countries.
Shadow of the Rock, 2s. 6d.
SHAFTESBURY. See English Philosophers.
SHAKESPEARE, ed. by R. G. White, 3 vols. 36s.;
édit. de luxe, 63s.
—— Annals; Life & Work, 2s.
—— Hamlet, 1603, also 1604, 7s. 6d.
—— Hamlet, by Karl Elze, 12s. 6d.
—— Heroines, by living painters, 105s.; artists'
proofs, 630s.
—— Macbeth, with etchings, 105s. and 52s. 6d.
—— Songs and Sonnets. See Choice Editions.
—— Taming of the Shrew, adapted for drawing-
room, paper wrapper, 1s.

SHEPHERD, British School of Painting, 2nd edit.

5s.; 3rd edit. sewed, 1s.
SHERIDAN, Rivals, col. plates, 52s. 6d. nett; art.
pr. 105s. nett.
SHIELDS, G. O., Big Game of North America, 21s.
—— Cruisings in the Cascades, 10s. 6d.
SHOCK, W. H., Steam Boilers, 73s. 6d.
SIDNEY. See Gentle Life Series.
Silent Hour. See Gentle Life Series.
SIMKIN, Our Armies, plates in imitation of water-
colour (5 parts at 1s.), 6s.
SIMSON, Ecuador and the Putumayor, 8s. 6d.
SKOTTOWE, Hanoverian Kings, new edit. 3s. 6d.
SLOANE, T. O., Home Experiments, 6s.
SMITH, HAMILTON, and LEGROS' French
Dictionary, 2 vols. 16s., 21s., and 22s.
SMITH, EDWARD, Cobbett, 2 vols. 24s.
—— G., Assyria, 18s.
—— Chaldean Account of Genesis, new edit. by
Sayce, 18s.
—— Gerard. See Illustrated Text Books.
—— T. Roger. See Illustrated Text Books.
Socrates. See Bayard Series.
SOMERSET, Our Village Life, 5s.
Spain. See Foreign Countries.
SPAYTH, Draught Player, new edit. 12s. 6d.
SPIERS, French Dictionary, 2 vols. 18s., half
bound, 2 vols., 21s.
SPRY. See Low's Stand. Library.

SPURGEON, C. H. See Preachers.

STANLEY, H. M., Congo, 2 vols. 42s. and 21s.
—— In Darkest Africa, 2 vols., 42s.
—— Emin's Rescue, 1s.
—— See also Low's Standard Library and Low's
Standard Books.
START, Exercises in Mensuration, 8d.
STEPHENS, F. G., Celebrated Flemish and French
Pictures, with notes, 28s.
—— See also Great Artists.
STERNE. See Bayard Series.
STERRY, J. Ashby, Cucumber Chronicles, 5s.
STEUART, J. A., Letters to Living Authors, new
edit. 2s. 6d.; édit. de luxe, 10s. 6d.
—— See also Low's Standard Novels.
STEVENS, J. W., Practical Workings of the Leather
Manufacture, illust. 18s.
—— T., Around the World on a Bicycle, over 100
illust. 16s.; part II. 16s.
STEWART, Dugald, Outlines of Moral Philosophy,
3s. 6d.
STOCKTON, F. R., Casting Away of Mrs. Leeks, 1s.
—— The Dusantes, a sequel, 1s.
—— Merry Chanter, 2s. 6d.
—— Personally Conducted, illust. by Joseph
Pennell, 7s. 6d.
—— Rudder Grangers Abroad, 2s. 6d.
—— Squirrel Inn, illust. 6s.
—— Story of Viteau, illust. 5s. new edit. 3s. 6d.
—— Three Burglars, 1s. & 2s.
—— See also Low's Standard Novels.

STORER, F. H., Agriculture, 2 vols., 25s.

STOWE, Edwin. See Great Artists.
—— Mrs., Flowers and Fruit from Her Writings, 3s.
6d.
—— Life ... her own Words ... Letters and Original
Composition, 15s.
—— Life, told for boys and girls, by S. A. Tooley,
5s., new edit. 2s. 6d. and 2s.
—— Little Foxes, cheap edit. 1s.; 4s. 6d.
—— Minister's Wooing, 1s.
—— Pearl of Orr's Island, 3s. 6d. and 1s.
—— Uncle Tom's Cabin, with 126 new illust. 2
vols. 18s.
—— See also Low's Standard Novels and Low's
Standard Series.
STRACHAN, J., New Guinea, 12s.
STRANAHAN, French Painting, 21s.
STRICKLAND, F., Engadine, new edit. 5s.
STUTFIELD, El Maghreb, ride through Morocco,
8s. 6d.
SUMNER, C., Memoir, new edit. 2 vols. 36s.
Sweden and Norway. See Foreign Countries.
Sylvanus Redivivus, 10s. 6d.
SZCZEPANSKI, Technical Literature, a directory,
2s.
TAINE, H. A., Origines, I. Ancient Régime, French
Revolution, 3 vols.; Modern Régime, vol. I. 16s.
TAYLOR, H., English Constitution, 18s.
—— R. L., Analysis Tables, 1s.
—— Chemistry, 1s. 6d.
Techno-Chemical Receipt Book, 10s. 6d.
TENNYSON. See Choice Editions.
Ten Years of a Sailor's Life, 7s. 6d.
THAUSING, Malt and Beer, 45s.
THEAKSTON, British Angling Flies, 5s.
Thomas à Kempis Birthday-Book, 3s. 6d.
—— Daily Text-Book, 2s. 6d.
—— See also Gentle Life Series.
THOMAS, Bertha, House on the Scar, Tale of South
Devon, 6s.
THOMSON, Joseph. See Low's Standard Library and
Low's Standard Novels.
—— W., Algebra, 5s.; without Answers, 4s. 6d.;
Key, 1s. 6d.
THORNTON, W. Pugin, Heads, and what they tell
us, 1s.
THORODSEN, J. P., Lad and Lass, 6s.
TICKNOR, G., Memoir, new edit., 2 vols. 21s.
TILESTON, Mary W., Daily Strength. 4s. 6d.
TINTORETTO. See Great Artists.
TITIAN. See Great Artists.
TODD, Life, by J. E. Todd, 12s.
TOURGEE. See Low's Standard Novels.
TOY, C. H., Judaism, 14s.
Tracks in Norway, 2s., n. ed. 1s.
TRAILL. See Prime Ministers.
Transactions of the Hong Kong Medical Society,
vol. I. 12s. 6d.
TROMHOLT, Aurora Borealis, 2 vols., 30s.
TUCKER, Eastern Europe, 15s.
TUCKERMAN, B., English Fiction, 8s. 6d.
—— Lafayette, 2 vols. 12s.
TURNER, J. M. W. See Gr. Artists.
TYSON, Arctic Adventures, 25s.
TYTLER, Sarah. See Low's Standard Novels.
—— M. C., American Literature, vols. I. and II.
24s.
UPTON, H., Dairy Farming, 2s.
Valley Council, by P. Clarke, 6s.
VANDYCK and HALS. See Great Artists.
VANE, Denzil, Lynn's Court Mystery, 1s.
—— See also Low's Standard Novels.
Vane, Young Sir Harry, 18s.
VELAZQUEZ. See Gr. Artists.
—— and MURILLO, by C. B. Curtis, with etchings,
31s. 6d. and 63s.
VERE, Sir F., Fighting Veres, 18s.
VERNE, J., Works by. See page 31.
Vernet and Delaroche. See Great Artists.
VERSCHUUR, G., At the Antipodes, 7s. 6d.
VIGNY, Cinq Mars, with etchings, 2 vols. 30s.
VINCENT, F., Through and through the Tropics,
10s. 6d.
—— Mrs. H., 40,000 Miles over Land and Water, 2
vols. 21s.; also 3s. 6d.
VIOLLET-LE-DUC, Architecture, 2 vols. 31s. 6d.
each.
WAGNER. See Gr. Musicians.
WALERY, Our Celebrities, vol. II. part i., 30s.
WALFORD, Mrs. L. B. See Low's Standard Novels.
WALL, Tombs of the Kings of England, 21s.
WALLACE, L., Ben Hur, 2s. 6d.
—— Boyhood of Christ, 15s.
—— See also Low's Stand. Novs.
WALLACE, R., Rural Economy of Australia and New
Zealand, illust. 21s. nett.