Journal of Soft Computing Paradigm (JSCP) (2020)

Vol.02/ No.04
Pages: 246-255
http://irojournals.com/jscp/
DOI: https://doi.org/10.36548/jscp.2020.4.006

Machine Learning Approach to Predictive

Maintenance in Manufacturing Industry
- A Comparative Study
Dr. P. Karrupusamy,
Professor and Head,
Department of Electrical and Electronics Engineering,
Shree Venkateshwara Hi-Tech Engineering College,
Erode, India.
Email: [email protected]

Abstract- Predictive maintenance is the way to improve asset management in every manufacturing industry. While handling
advance costlier machinery in the industry, the predictive maintenance knowledge will be essential to protect the machinery
before gets degradation performance. Recently, the emergence of business in manufacturing industry deals with good systems,
regular intervals maintenance process, predictive maintenance (PdM), machine learning (ML) approaches are extensively applied
for handling the health standing of business instrumentation. Now the digital transformation towards I4.0, data techniques,
processed management and communication networks; it’s doable to gather huge amounts of operational and processes conditions
information generated type many items of kit and harvest information for creating an automatic fault detection and diagnosing
with the aim to attenuate period of time and increase utilization rate of the parts and increase their remaining helpful lives. The
predictive maintenance is inevitable for property good producing in I40. This paper aims to provide a comprehensive review of
the recent advancements of metric capacity unit techniques wide applied to PdM for good producing in I4.0 by classifying the
analysis consistent with metric capacity unit algorithms, ML class, machinery and instrumentation used device employed in
information acquisition, classification of knowledge size and kind, and highlight the key contributions of the researchers and so
offers pointers and foundation for additional analysis. In this research paper we constructed a Random Forest model to predict the
failure of the various machine in manufacturing industry. It compares the prediction result with Decision Tree (DT) algorithm
and proves its superiority in accuracy and precision.

Keywords: Industry 4.0, Analytic Models, Machine Learning, IIoT Sensors

1. INTRODUCTION
The predictive maintenance is the future scope of various industries for many reasons. The term “Industry
4.0 (or) I4.0” is deriving from “The Fourth Industrial Revolution” currently famous word among industries territory
[1]. It is necessary to handling the industrial equipment very safely and it should be healthier in their status always.
If they maintain the status is healthier, then they can increase manufacturing profit into double. The machine
learning techniques have emerged as a promising tool in PdM application for good manufacturing in I4.0; so it’s
exaggerated attraction of authors/researchers throughout recent years [2]. Simple manufacturing company consist of
various operation and process with cutting machine, induction motor with pumping system, refrigeration system,
and little power generation unit by wind turbine and etc. The importance of predictive maintenance is that the
industry consists of physical and digital system for the any production context currently. Also the selection of
inappropriate predictive maintenance technique from wrong dataset and it leads to infeasible maintenance
scheduling [3]. The Machine learning (ML) algorithm is mainly suitable to detect fault with the help of huge
collected dataset. But its bit challenge task to select appropriate ML techniques in an industrial systems. The older
lean management was limited problem finder for today’s companies also it is struggling to drive better efficiency
currently [4].

As shown in figure 1, they will fix the problem or defect once it breaks in ancient times. Then they
maintain the machine with regular intervals checking. After that, predict the exactly when it will break and maintain
accordingly with historical data. Recently, the machine itself decides how to avoid and providing warning symbol or
alarm for predicted imminent failure in the machine with learning algorithm [4, 5].

246
ISSN: 2582-2640 (online)
Submitted: 20.12.2020
Accepted: 19. 01.2021
Published: 27.01.2021
Journal of Soft Computing Paradigm (JSCP) (2020)
Vol.02/ No.04
Pages: 246-255
http://irojournals.com/jscp/
DOI: https://doi.org/10.36548/jscp.2020.4.006

Figure 1 The evolution of maintenance strategies

Generally, the predictive maintenance is based on sensor measurement for e.g. temperature, pressure,
vibration, rotation speeds, current, chemical properties of oil or water. Study of behavior of the system, current
condition of the system, which parts of the machine are mostly failed many times, what will changes in the system
due to temperature, vibration, pressure and etc. Not only these above said questions and many more intelligent
version questions are better analytic results for predictive maintenance for machine [6, 7].

PREDICTION

PROACTIVE Maintenance
SCHEDULED
REACTIVE

Figure 2 The stepping level of predictive maintenance

The figure 2 shows the level 4 can protect unpredictable failures more rather than previous level set of
dataset. So the machine learning algorithms is essential tool to predict the efficient maintenance for machine now a
day. During construction, catastrophic failures is one of the error which recovery impossible. The predictive
maintenance is the only solution for this issue. The successful predictive maintenance systems are ensuring good
accuracy and validity [8]. The maintenance of sensors, interface units (display, monitoring, and alarm) will be more
constructive and systematic approach too. This predictive measure can create new era dashboard and indicators to
run the manufacturing business in well manner [9, 10].

The vertical axis in the graphs shows reliability of the methods. The reliability is increasing technique by
technique over the period. The horizontal axis in the graphs shows evaluation years. In an industry, the machines are
repaired by reactive methods. The machine will be fixed after breakdown. So the overall product will be decreased.
The various machines is having several type of achievement and work nature. Then they started to do scheduled
maintenance activities. This method also wastes of human resources and time. Proactive method also lagging and
fails in overall predicting the maintenance timing for any machine. Finally, predictive maintenance is an advanced
analytics and sensing the data to detect the machine reliability.

The benefits of predictive maintenance

Increasing machine useful life
Due to set operation in an optimum setting, the machine breakdown reduces; causes it increases machine
life. One of the stressful failures is asset failure in the system level. Due to operation reliable, the downtime of
machine or system will be reduced.

247
ISSN: 2582-2640 (online)
Submitted: 20.12.2020
Accepted: 19. 01.2021
Published: 27.01.2021
Journal of Soft Computing Paradigm (JSCP) (2020)
Vol.02/ No.04
Pages: 246-255
http://irojournals.com/jscp/
DOI: https://doi.org/10.36548/jscp.2020.4.006

Reduced environment impact

With the aid of collection of advance analytics, the machine operates for longer period so the system
consumes fewer natural resources. Predictive maintenance is handling social responsibility very caring for the
environment impact.

Reduction of maintenance costs

The sensor operations are very perfect and very efficient due to planning and scheduled maintenance
provides less maintenance cost.

Reduction in workers injuries

Obviously, the reduced breakdowns will help to predict the sensor data perfectly which reduces worker
injuries rate.

Reduction in waste work in the system

If the optimum operation is not happened for the system, the wasteful production will be a result or output.
But predictive maintenance system can cover optimum operation due to advance analytics.

Improved product quality

The sensor data will be evaluated correctly by input algorithm can create and improve the product quality
for longer time and get customer’s worthy feedback.

Increased performance over time

The predictive maintenance systems are operating based on knowledge or intelligent based learning
systems. So, the decision making technique will improve overall performance on over time implicit. The predictive
maintenance applications are implementing on automotive, airlines, high-tech manufacturing, transportation, harbor
port that industries are gaining those benefits directly. These industries are directly related or dealing with
monitoring sensor data [11, 12].

There are lots of advantages of predictive maintenance from cost saving sector and it includes minimizing
planned downtime of the system lifespan. Also, the amount of time it takes to assess and implement a predictive
maintenance schedule.

2. ORGANIZATION OF THE RESEARCH

The structure of the research article organized as follows; Section 3 gives literature survey of recent predictive
maintenance program and its approach. Section 4 provides the description of theoretical analysis. Section 5 delivers
description of results and discussion finally the conclusion and further improvement is in section 6.

3. PRELIMINARIES
Sathish & Smys says about smart agriculture, living, parking, cities, industry as well as environment. The
industry consists of explosive type of gas and material which should be monitored and controlled very carefully.
This maintenance and repair should be in early and predictive maintenance assigned for a particular failure which is
not reversible [1]. Farahani & Firouzi et al proposes use of machine learning analytics with huge amount of data
from the Internet of Things (IoT) sensors and it is controlled distributed the application into various manufacturing
industry sector. They mainly focuses predict the failure in the system in advance, therefore increasing machine
useful lifespan. Also they indicate about predictive maintenance research is in still preliminary level in
manufacturing industry [13]. Chiang & Zhang concentrates oil and gas industry in manufacturing unit [14].

Lee, Ghaffari & Elmeligy research focuses a Condition Based Maintenance (CBM) strategy which is in
earlier techniques for predictive method. Previous revolution in industry mostly, the system used to detect any
failures beforehand with the aid of Preventive Maintenance (PM) strategies [15]. Selcuk research gave that the
importance of Predictive Maintenance in terms of improving reliability, efficiency and prefer green environment
[16]. Civerchia et al introduces system and component architecture of Industry Internet of Things (IIoT) sensor and
it is detailed described [17]. Mourtzis et al discuss about predictive Maintenance and IoT applications with IIoT

248
ISSN: 2582-2640 (online)
Submitted: 20.12.2020
Accepted: 19. 01.2021
Published: 27.01.2021
Journal of Soft Computing Paradigm (JSCP) (2020)
Vol.02/ No.04
Pages: 246-255
http://irojournals.com/jscp/
DOI: https://doi.org/10.36548/jscp.2020.4.006

sensor architecture. The work focuses IoT wireless sensor network is working for industrial applications which
maintains with huge volume of dataset for efficient analysis about the machine [18]. Wang proposes zero defect
manufacturing in an industry with the aid of predictive maintenance technique. Moreover, the author discuss about
predictive maintenance consists of number of activities which will be in the dataset. This can help to detect the
changes in physical state of machine that choose appropriate maintenance work for equipment failure. This will lead
to increase the performance over time. This allotment of appropriate maintenance work is in view of intelligent
predictive Maintenance framework. Mainly, the author focuses an optimal prediction of machinery failures using
huge volume of dataset from IIoT sensors [19].

The above said research articles are perfect class in proactive, scheduled maintenance analysis. Also the
algorithm is very inefficient in predictive maintenance schedule. There is research gap being there that lack of this
machine predicting its failure and protecting. We developed a model and conducted several test with the help of
learning algorithm for industrial machinery. The main goal of this research study is to discover defect in the various
machine parts with the ML techniques in an industry.

4. THEORETICAL ANALYSES
The cloud computing will be more smarter while using machine learning techniques in that. The machine
learning techniques are very simple and popular to handle the data in cloud. In this research paper focuses AutoML
(automated Machine Learning) for the process in real and practical environment. The AutoML consist of two main
aspects named data collection and prediction. The machine process will be optimized due to AutoML after
parameter tuning. Here automated feature detection and selection has happening after the data preprocessed from
dataset. But each step is controlled and performing manually with the reference from previous one [20, 21, 22].
Currently, the ML technique has operated in wider range application in manufacturing industry. Because of this ML
algorithm provide better support for any kind of manufacturing industry always. Obviously, we can reduce part of
manpower in any industry with increasing efficiency. In this research papers, the Bayes naïve algorithm is used in
parameter tuning process to give efficient computing power.

4.1 Important Units

IIoT Sensors
Many sensors are operated as IIoT which is used for interact and communicate with cyber physical
systems. Some of the sensors measures temperature, pressure, electrical quantities, vibration, and humidity. The
output of sensors can be conditioned by signal conditioner unit in our proposed model. The more number of IIoT
sensors uses for many varieties of application targets such as air, gas, any liquid quality measures; also ultrasonic
sound sensor is used to measure many leakages in tank which is in industry for many purposes [23, 24].

Signal Conditioning Unit

Generally, this unit is making a signal in appropriate format for next section which will connect with signal
conditioning unit. The capturing signals are taken as data and convert into binary format using Analog-Digital
Converter (ADC). For further analysis, the signals should be in digital format. It is shown in the proposed system
architecture in the figure 5.

Data Creation & Communication

There must be a database file which is remotely centralized. Because of the IoT sensors are located
different locations and installed in remote location without internet connection.

Extension of cloud computing

The predictive maintenance is computed by edge or cloud. The edge computing is local network level
which is used to provide real time machine failure. Normally, the cloud computing is having big data center that is
located in centralized. The IoT sensors are providing data for further processing to clouds. Usually, the latency
problems arise in cloud computing which is mitigated by fog architecture implementation over the cloud which is
shown in figure 3.

249
ISSN: 2582-2640 (online)
Submitted: 20.12.2020
Accepted: 19. 01.2021
Published: 27.01.2021
Journal of Soft Computing Paradigm (JSCP) (2020)
Vol.02/ No.04
Pages: 246-255
http://irojournals.com/jscp/
DOI: https://doi.org/10.36548/jscp.2020.4.006

Labeled Observation
The labeled data is having form of training set and test set for the classification by machine learning
algorithm. Based on the error, the labeled data will be countable and give feedback for the prediction process [25,
26, 27, 28].

Figure 3 Fog computing between cloud and end devices

4.2 Proposed Predictive Maintenance model

4.2.1. Decision Tree (DT)
Decision Tree is a network devices that consists of branch which is connected many nodes as shown in
figure 4. The nodes are classified into two types.

1. Root nodes
2. Intermediate nodes

Figure 4 Overview of DT based Predictive Maintenance

The root nodes are used as backbone of the network system. The intermediate nodes are used to connect
between root nodes and leaf nodes. This type of nodes is used to represent the features. Also it is named as decision
nodes.

250
ISSN: 2582-2640 (online)
Submitted: 20.12.2020
Accepted: 19. 01.2021
Published: 27.01.2021
Journal of Soft Computing Paradigm (JSCP) (2020)
Vol.02/ No.04
Pages: 246-255
http://irojournals.com/jscp/
DOI: https://doi.org/10.36548/jscp.2020.4.006

Signal Conditioning Data Creation &

IIoT Sensors
Circuit Communication

Extension of Cloud
Computing

Labeled Observation

Training Dataset Testing Dataset

Machine Learner
Prediction Model Statistics

Figure 5 Overview of system architecture for Predictive Maintenance

4.2.2. Random Forest (RF)

This algorithm is one of the better accuracy wise result algorithms among supervised classification
algorithm in an industry sector computation. This algorithm used to create forest and select it as random from that.
The result will be more accurate while it contains larger number of trees. The Random Forest algorithm is used to
find the root node and splitting the feature nodes randomly. Also this algorithm comprises of several DT classifiers
and it contains many nodes. This prediction always based on tree predictions and our prediction RF model is
constructed as shown in figure 6.

Our study is focusing the predicting the machine’s imminent failure in the real-time in an industry with the
help of Random Forest (RF model. There are number of metrics analyzed to compute the algorithm capability. Many
ML algorithms can predict the healthier of the machine. It includes Linear Regression (LR), RF and Symbolic
Regression (SR).

251
ISSN: 2582-2640 (online)
Submitted: 20.12.2020
Accepted: 19. 01.2021
Published: 27.01.2021
Journal of Soft Computing Paradigm (JSCP) (2020)
Vol.02/ No.04
Pages: 246-255
http://irojournals.com/jscp/
DOI: https://doi.org/10.36548/jscp.2020.4.006

Figure 6 Random Forest machine learning algorithm

There are many detecting methods that drifting behavior in the data stream, lack of continues data stream,
variation in vibration measurement and thermal imaging data, abrupt changes in sensor data in continues monitoring
sensor streams. From the real data, our consideration is that rotating machinery in an industry. The RF algorithm is
used as the multi sensor system for rotating machinery using feature fusion method. Also this technique can
compensate the short comings of the heat or vibrations. It will provide the significant changes in the machine or
sensor performance also its detection unit. For the industry pumps, the RF algorithm considering real data type is
used to predict the failure of it. Also this can detect faults in vibration data in the machine data sets. The proposed
model was validated using from few months monitoring data sets of the machine. For the cutting machine in the
industry, model can evaluate its vibration measurement and thermal imaging data of it. Also it provides high
accuracy result in final. For all type of refrigeration systems, monitoring the temperature sensor changes and it is
validated with real data.

5. RESULT DISCUSSIONS
The obtained result has shown and discuss in table 1. The rotating machines are evaluated by the feature
fusion measurement for fault detection. This technique provides the result based on vibration data driven features.
These extracted features are providing fault detection with moderate accuracy from an individual sensor streams.
The transient current signal are evaluating for the induction motor status. These features are extracted and to
determine the fitness of the machine. Also this double classifier approach is preferred for detect fault in the machine
with the help of current and voltage waveform characteristic.

252
ISSN: 2582-2640 (online)
Submitted: 20.12.2020
Accepted: 19. 01.2021
Published: 27.01.2021
Journal of Soft Computing Paradigm (JSCP) (2020)
Vol.02/ No.04
Pages: 246-255
http://irojournals.com/jscp/
DOI: https://doi.org/10.36548/jscp.2020.4.006

Table 1 The summary of obtained results

The refrigeration systems are in many supermarket as well as chemical industry. In order to predict the
fault occurrence in the machine, the algorithm can detect the fault early and validated with real time data with 2125
different refrigerator. The model achieved pre identification and good precision, accuracy. It will find more
beneficial such as preventive maintenance and proactive scenario for many industries.

100
80
60
40
20
0 Decision Tree
Random Forest

Figure 7 Accuracy evaluation chart

From the industrial cutting machinery dataset, the algorithm model can predict exact faultiness of the
machine due to high amount of features from the dataset. The industrial pump can predict faultiness from sensor
vibration data and provide high accuracy nearly. Moreover, the decision tree algorithm model is lesser in all aspects

253
ISSN: 2582-2640 (online)
Submitted: 20.12.2020
Accepted: 19. 01.2021
Published: 27.01.2021
Journal of Soft Computing Paradigm (JSCP) (2020)
Vol.02/ No.04
Pages: 246-255
http://irojournals.com/jscp/
DOI: https://doi.org/10.36548/jscp.2020.4.006

compared to RF model algorithm. DT model is an unstable that cannot withstand small changes in dataset which is
acquired from an industry. They are relatively inaccurate with prediction during changes in dataset in real-time. It
shows in the graph figure 7.

6. CONCLUSION
This research article present a review of RF and DT methods of machine learning algorithm applies in
various parts of manufacturing industry. Finally constructed RF and tested successfully with various metric
measurements. The discussion and literature review also was performed. The determining machine failure is a
primary goal of predictive maintenance. All the sensors data sent and processing by learning algorithm for predict
the machine failure. The given data is preprocessed because of the real time data. These data is collected directly
from sensor with raw. So the data are not clean and ordered. The cleaning data carried out for further process which
provides higher reliability and accuracy of prediction. Also this data are splitting into two for training and testing.
With the help of these dataset, there can create predictive maintenance model. The output from the constructed
model will be analyzed and evaluated. The accuracy and precision will be fewer in some period of predictive time.
The deep logical analysis of the key demanding situations is challenging task as our future enhancement of our
research article. Also the various segments in manufacturing industry and minimum predictive maintenance data
size will provide less accuracy and precise for predictive maintenance.

Also we recommend do furthering improvements and motivating authors and researchers for the following aspects;
1. The optimum preprocessing technique should be used for raw data analysis to get better accuracy.
2. Combine two or more ML techniques to design the model to achieve better prediction for data acquisition.
3. Machine Learning model approach can be extended for further studied.
4. Classification and anomaly detection algorithms will be combined to keep up exactness of the classification
models while not losing anomaly detection benefits. By this manner, PdM can be applied to instrumentality
or system that doesn’t have giant data.

ACKNOWLEDGEMENT
We are most grateful to Eng. Mussie, (Retired) Ministry of Telecommunication, Department of Electrical
& Electronics Engineering, Mainefhi College of Engineering, for his assistance and encourage this research.

REFERENCES
[1] Sathish and Smys “A Survey on Internet of Things (IoT) based Smart Systems” Journal of ISMAC (2020)
Vol.02/ No.04 Pages: 181-189, http://irojournals.com/iroismac/ DOI: https://doi.org/10.36548/jismac.2020.4.001
[2] Carvalho, T.P.; Soares, F.A.; Vita, R.; Francisco, R.D.; Basto, J.P.; Alcalá, S.G. A systematic literature review of
machine learning methods applied to predictive maintenance. Comput. Ind. Eng. 2019, 137, 106024.
[3] Cinar, Z.M.; Nuhu, A.A.; Zeeshan, Q.; Korhan, O. Digital Twins for Industry 4.0: A Review. In Industrial
Engineering in the Digital Disruption Era. GJCIE 2019. Lecture Notes in Management and Industrial Engineering;
Calisir, F., Korhan, O., Eds.; Springer: Cham, Switzerland, 2020.
[4] Zhang, W.; Yang, D.; Wang, H. Data-Driven Methods for Predictive Maintenance of Industrial Equipment: A
Survey. IEEE Syst. J. 2019, 13, 2213–2227.
[5] Borgi, T.; Hidri, A.; Neef, B.; Naceur, M.S. Data analytics for predictive maintenance of industrial robots. In
Proceedings of the 2017 International Conference on Advanced Systems and Electric Technologies (IC_ASET),
Hammamet, Tunisia, 14–17 January 2017; pp. 412–417.
[6] Dai, X.; Gao, Z. From model, signal to knowledge: A data-driven perspective of fault detection and diagnosis.
IEEE Trans. Ind. Inform. 2013, 9, 2226–2238.
[7] Lee, J.; Lapira, E.; Bagheri, B.; Kao, H.A. Recent advances and trends in predictive manufacturing systems in
big data environment. Manuf. Lett. 2013, 1, 38–41.
[8] Peres, R.S.; Rocha, A.D.; Leitao, P.; Barata, J. IDARTS–Towards intelligent data analysis and real-time
supervision for industry 4.0. Comput. Ind. 2018, 101, 138–146.
[9] Sezer, E.; Romero, D.; Guedea, F.; MacChi, M.; Emmanouilidis, C. An Industry 4.0-Enabled Low Cost
Predictive Maintenance Approach for SMEs. In Proceedings of the 2018 IEEE International Conference on
Engineering, Technology and Innovation (ICE/ITMC), Stuttgart, Germany, 17–20 June 2018; pp. 1–8.

254
ISSN: 2582-2640 (online)
Submitted: 20.12.2020
Accepted: 19. 01.2021
Published: 27.01.2021
Journal of Soft Computing Paradigm (JSCP) (2020)
Vol.02/ No.04
Pages: 246-255
http://irojournals.com/jscp/
DOI: https://doi.org/10.36548/jscp.2020.4.006

[10] Biswal, S.; Sabareesh, G.R. Design and development of a wind turbine test rig for condition monitoring studies.
In Proceedings of the 2015 International Conference on Industrial Instrumentation and Control (ICIC), Pune, India,
28–30 May 2015; pp. 891–896.
[11] Amruthnath, N.; Gupta, T. A research study on unsupervised machine learning algorithms for early fault
detection in predictive maintenance. In Proceedings of the 2018 5th International Conference on Industrial
Engineering and Applications, ICIEA, Singapore, 26–28 April 2018; pp. 355–361.
[12] Yoo, Y.; Park, S.H.; Baek, J.G. A Clustering-Based Equipment Condition Model of Chemical Vapor
Deposition Process. Int. J. Precis. Eng. Manuf. 2019, 20, 1677–1689.
[13] Farahani, B., Firouzi, F., Chang, V., Badaroglu, M., Constant, N., & Mankodiya, K. (2018). Towards fog-
driven IoT eHealth: Promises and challenges of IoT in medicine and healthcare. Future Generation Computer
Systems, 78, 659-676.
[14] Chiang, M., & Zhang, T. (2016). Fog and IoT: An overview of research opportunities. IEEE Internet of Things
Journal, 3(6), 854-864.
[15] Lee, J., Ghaffari, M., & Elmeligy, S. (2011). Self-maintenance and engineering immune systems: Towards
smarter machines and manufacturing systems. Annual Reviews in Control, 35(1), 111-122.
[16] Selcuk, S. (2017). Predictive maintenance, its implementation and latest trends. Proceedings of the Institution
of Mechanical Engineers, Part B: Journal of Engineering Manufacture, 231(9), 1670-1679.
[17] Civerchia, F., Bocchino, S., Salvadori, C., Rossi, E., Maggiani, L., & Petracca, M. (2017). Industrial Internet of
Things monitoring solution for advanced predictive maintenance applications. Journal of Industrial Information
Integration, 7, 4-12.
[18] Mourtzis, D., Vlachou, E., & Milas, N. (2016). Industrial Big Data as a result of IoT adoption in manufacturing.
Procedia cirp, 55, 290-295.
[19] Wang, K. (2016). Intelligent predictive maintenance (IPdM) system–Industry 4.0 scenario. WIT Transactions
on Engineering Sciences, 113, 259-268.
[20] Susto, G.A.; Schirru, A.; Pampuri, S.; McLoone, S.; Beghi, A. Machine learning for predictive maintenance: A
multiple classifier approach. IEEE Trans. Ind. Inform. 2015, 11, 812–820.
[21] Susto, G.A.; Beghi, A.; De Luca, C. A predictive maintenance system for epitaxy processes based on filtering
and prediction techniques. IEEE Trans. Semicond. Manuf. 2012, 25, 638–649.
[22] Ahmad, R.; Kamaruddin, S. An overview of time-based and condition-based maintenance in industrial
application. Comput. Ind. Eng. 2012, 63, 135–149.
[23] Mathew, V.; Toby, T.; Singh, V.; Rao, B.M.; Kumar, M.G. Prediction of Remaining Useful Lifetime (RUL) of
turbofan engine using machine learning. In Proceedings of the 2017 IEEE International Conference on Circuits and
Systems (ICCS), Thiruvananthapuram, India, 20–21 December 2017; pp. 306–311.
[24] Samuel, A.L. Some studies in machine learning using the game of checkers. IBM J. Res. Dev. 2000, 44, 206–
226.
[25] Rivera Torres, P.J.; Serrano Mercado, E.I.; Llanes Santiago, O.; Anido Rifón, L. Modeling preventive
maintenance of manufacturing processes with probabilistic Boolean networks with interventions. J. Intell. Manuf.
2018, 29, 1941–1952.
[26] Jezzini, A.; Ayache, M.; Elkhansa, L.; Makki, B.; Zein, M. E_ects of predictive maintenance(PdM), Proactive
maintenace(PoM) & Preventive maintenance(PM) on minimizing the faults in medical instruments. In Proceedings
of the 2013 2nd International Conference on Advances in Biomedical Engineering, Tripoli, Lebanon, 11–13
September 2013; pp. 53–56.
[27] Kumar, A.; Chinnam, R.B.; Tseng, F. An HMM and polynomial regression based approach for remaining
useful life and health state estimation of cutting tools. Comput. Ind. Eng. 2019, 128, 1008–1014.
[28] Munirathinam, S.; Ramadoss, B. Big data predictive analtyics for proactive semiconductor equipment
maintenance. In Proceedings of the 2014 IEEE International Conference on Big Data (Big Data), Washington, DC,
USA, 27–30 October 2014; pp. 893–902.

255
ISSN: 2582-2640 (online)
Submitted: 20.12.2020
Accepted: 19. 01.2021
Published: 27.01.2021