Prediction of Groundwater Quality Using Artificial Neural Network
Assessment and prediction of water quality is a vital tool for the management of water resources systems. It is necessitous for human existence, agriculture and industry. This project delves into the prediction of groundwater quality parameters and groundwater quality criterion based on the Artificial Neural Network Modelling with the study area as Kerala, a state of India. Two models were developed. The first model employs the water quality parameters such as pH, electrical conductivity, total hardness as the input parameters and calcium, magnesium, chloride, fluoride, nitrate concentration as the output parameters. The second model was designed by giving input as, input values and the predicted output values of the first model, and groundwater quality criterion corresponding to each location as the target values. The output qualitative parameters were estimated and compared with the measured values, to evaluate the influence of key input parameters. The number of neurons to be given in the hidden layer was decided by the trial-and-error method. Data of 506 water samples from all over Kerala were collected for modelling. The results show that the performance of the first model is having a regression value R2=0.97 and the second model is having a regression value R2=0.99, using backpropagation algorithms according to the best-chosen input parameters. MATLAB R2018a was used for developing the multi-layer perceptron ANN models and the performance function for calculating the model performance error statistics was the coefficient of determination (R2). This project introduces a cost-effective and quick method for the estimation of groundwater quality.
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.