Estimation of water quality index using modern-day machine learning algorithms

Abstract

Many human-made activities currently pollute groundwater supplies, with mining operations playing a substantial role in this degradation. In this study, water quality index (WQI) was calculated and forecasted for groundwater in gold mining sites of Kolar Gold Fields, Karnataka, using several water quality criteria and modern-day soft computing approaches. Specifically, three sophisticated deep learning models: convolution neural network (CNN), deep neural network (DNN), and recurrent neural network were used to estimate the WQI using various water quality metrics. The outcomes of these models were also compared with three widely used soft computing models namely support vector machine (SVM), least-square support vector machine (LS-SVM), and artificial neural network. Experimental results reveals that the developed CNN model outperform other two models with R² values of 0.9998 and 0.9996 in the training and testing phases, respectively. The RMSE values of the CNN model were determined to be 0.0034 and 0.0038 in the training and testing phases, respectively. As per the results, the developed CNN model can be used as alternate tool for rapid water quality monitoring.