Quality Prediction in a Mining Process

Lokesha E J Punith B N Abhishek

Computer Science and Engineering Computer Science and Engineering Computer Science and Engineering
BIET BIET BIET
Davangere, India Davangere, India Davangere, India
[email protected] [email protected] [email protected]

Aman Prof. S B Mallikarjuna Dr.Nirmala C R

Computer Science and Engineering Computer Science and Engineering Computer Science and Engineering
BIET BIET BIET
Davangere, India Davangere, India Davangere, India
[email protected] [email protected] [email protected]

Abstract— Froth flotation is a process that is used to model for prediction percentage of silica concentrate with and
separate the materials hydrophobic and hydrophilic
minerals , for the quality prediction of mining process is without ore concentrate in the froth flotation by using random
removal of impurities within the ore concentration in the forest regression algorithm.
lab experiment it will take at least one hour for engineers
for prediction of impurity so that for every hour
Artificial Intelligence and Machine Learning
engineers will predict the impurity it will be time
consuming so that we proposed a model for engineers to
help for the prediction of with iron ore and without iron Artificial intelligence (AI) is the demonstrates of human
ore concentrate so that engineers will take the corrective intelligence processes by machines and feeding the experience
actions in the early as possible.
of human to machines in the form of data, greatly help in Qualti
Keywords: - forth flotation, minerals, slica, iron ore
prediction in a Mining Process.
I. INTRODUCTION

In iron ore Mining industry in order to specified II. Literature Survey

quality within the froth flotation process plant.
1] Author in the paradigm of froth flotation system has explored
engineers admit typical laboratory to check the and discovered the most significant parameters that influence
technique for the check impurities, it will take quite the flotation performance of lead mineral. Seemingly, the new
model suggested that grinding time, flotation pH, for
two hours to establish the process of flotation in mining
comparable collector, solid-in-pulp concentration and the
process, and how to increase the quantity slica impurity increase of solid-in-pulp concentration have the most significant
with the extracted of iron ore, silica to increase the effect on the ore recovery and selective separation of lead
mineral. He concluded that solid-in-pulp concentration was the
quantity of iron within the extracted ore, chemical most important parameter that influences the flotation of lead
reagents are more added into the ore pulp so that mineral [4].
modify and removal of impurities this method is that 2] However, Bergh el [5] all conducted a pilot and industrial
the targeted ore quantitative prediction of the quality. research to establish how the characteristics of flotation
processes, the quality of measurements of key variables and
the main goal is to predict the percentage of silica
general lack of robust models are thwarting the appropriate use
concentration of the froth flotation process in plant by of predictive control. The authors proposed a multivariate
using machine learning techniques for the proposed statistics model such as PCA to explain the relationship
between operation data for on-line diagnosis and fault B. Importing Libraries
detection. It was found that statistical methods seem to
provide a general framework to build models in latent To perform data preprocessing in using Python, we need to
variables related to froth characteristics at short import some predefined Python libraries. These libraries are
sampling internal. The authors further discovered a used to perform jobs.
supervisory and stabilizing control is a form of sub-
optimal expert systems. The only challenge found was
the difficulties in replicating a particular solution from C. Split the Dataset into the Training set and Testing set
one plant to another [5].
Our dataset is divided into a train set and test set. by doing this
we can increase the performance of our machine learning
model.
III. DATASET COLLECTION
Train set is used to train our model and Test set used to check
Dataset: weather our model working correct or not
https://www.kaggle.com/edumagalhaes/quality-
prediction-in-a-mining-process

We are collected the dataset from kaggle, kaggle is a

online community for the collection of various research
fields datasets from data analytic practitioners. in our
dataset contain 24 features, Our model can be pre
trained with certain features like date and time, amina Fig. 2.Train and Test Data Split
flow, froth flotation air level etc..
D. Building the Model

After the completion of data pre-processing &training and

IV. METHODOLOGY testing of dataset then we have to build the model by using
algorithms

Decision trees are fall under the category of supervised

machine learning means decision trees construct the model for
prediction /decision of given record. Decision tress are fast and
accurate

Linear regression is used the finding the co-relation b/w the

independent variables and dependent variables

Random forest regression is the combination of many decision

trees means it takes average of the all decision trees. random
forest is to perform both classification and regression & it
improves the accuracy of the model and prevents the over
fitting issues.

Fig. 1. Methodology For Prediction Crops

A. Data Preprocessing

A real-world data typically contains noises, missing

values, and errors in which cannot be directly used for
the machine learning models. Data preprocessing is
required for the cleaning the data and making it suitable
for a machine learning model which also increases
efficiency of a machine learning model and the
accuracy
Fig. 3. Dataset Visualisation
 V. RESULTS

Mean squared error (MSE) is the average of sum of

squared difference between actual value and the
predicted or estimated value. It is also termed as mean
squared deviation (MSD). This is how it is represented
mathematically:

Fig. 4. Mean squared error

R-Squared is the ratio of Sum of Squares Regression

(SSR) and Sum of Squares Total (SST). Sum of
Squares Regression is amount of variance explained by
the regression line. R-squared value is used to measure
the goodness of fit

Fig. 6. Home Page

Fig. 5. R-Squared

R2 (coefficient of determination) regression score

function.

Best possible score is 1.0 and it can be negative

(because the model can be arbitrarily worset). A
constant model that always predicts the expected value
of y, disregarding the input features, would get
a R2 score of 0.0

TABLE I: SUMMARY OF RESULTS

Attribute MSE RMSE R2_SCORE

Silca 0.00245 0.0495 0.9941
impurity
prediction
without Fig. 7:- To Predict silica (impurity) % with Iron Ore
iron ore
Silca 0.0438 0.00192 0.99391
impurity
prediction
with iron
ore
 REFERENCES

[1] https://www.diva-
portal.org/smash/get/diva2:1386720/FULLTEXT01.pdf
[2] https://www.kaggle.com/edumagalhaes/quality-prediction-
in-a-mining-process
[3] https://medium.com/analytics-vidhya/using-a-lstm-neural-
network-to-predict-a-mining-industry-process-parameter-
d91df7ffb4e
[4] Luo, X., Feng, B., Wong, C., Miao, J., Ma, B., & Zhou, H.
(2016). The critical importance of pulp concentration on
the flotation of galena from a low grade lead–zinc ore.
Journal of Materials Research and Technology, 5(2), 131–
135.
[5] Bergh, L. G., & Yianatos, J. B. (2011). The long way
toward multivariate predictive control of flotation
Fig. 8. To Predict silica (impurity) % without Iron Ore
processes. Journal of Process Control, 21(2), 226–234.

VI. CONCLUSIONS [6] https://www.kaggle.com/edumagalhaes/quality-prediction-in-

a-mining-process
The introduction of contemporary technologies like [7] Michaud, D. (2017). Froth Flotation Process. Retrieved
September 23, 2019, from
cloud computing, artificial intelligence and machine
[8] Mineral Processing & Metallurgy website:
learning and these technologies are used in a different https://www.911metallurgist.com/blog/froth-
area such as soil classification, disease identification [9] flotation-process
and classification and automated irrigation system and [10] Nakhaei, F., &Irannajad, M. (2015). Application and
also going to be helpful for prediction of the crop. this comparison of RNN, RBFNN and
paper includes an overview of machine learning [11] MNLR approaches on prediction of flotation column
approach for prediction of crops & fruits like wheat, performance. International Journal of
[12] Mining Science and Technology, 25(6), 983–990.
mango, rice, coconut etc. . We proposed the model that
[13] Chen, W., Tai, P., Wang, M., Deng, W., and Chen, C. 2008. A
predicts a best suitable for the crop/fruits for land Neural Network-Based Approach for Dynamic Quality
which gives the high yield production based on a Prediction in A Plastic Injection Molding Process. Expert
machine learning approach, our model used the concept Systems with Applications. 35 (3), 843–849.
of random forest regression & we also observe random [14] Dawson, P., &Koorts, R. (2014). Flotation Control
forest regression algorithm which gives Incorporating Fuzzy Logic and
results/accuracy is high compare to other machine [15] Image Analysis: IFAC Proceedings Volumes, 47(3), 352–357
learning algorithms.