ACCURATE PREDICTION OF WATER QUALITY INDEX USING ARTIFICIAL NEURAL NETWORK MODELING IN THE SOUTH-WESTERN REGION OF VIZIANAGARAM DISTRICT, ANDHRA PRADESH, INDIA
Groundwater quality deterioration is a growing concern in rural India, where aquifers serve as the primary drinkingwater source. This study evaluates and predicts the quality of groundwater in the selected villages of the South- Western region of Vizianagaram District, Andhra Pradesh, using a Water Quality Index (WQI) integratedwithArtificial Neural Network (ANN) modelling. A total of 864 Groundwater samples from 24 locations were collectedduring 2018–2021 and analysed for fifteen physicochemical parameters following APHA procedures and calculatedWQI using the weighted Arithmetic method. WQI classification indicated 58.33% of samples fall under goodquality, 29.17%poor, 4.17% very poor, and 8.33% unsuitable for drinking. To minimise the dependence onextensive laboratory analyses, A feed-forward back-propagation ANN model with architecture 15–5–1 trained usingthe Levenberg–Marquardt algorithm was developed and obtained excellent predictive performance (R2=0.999, RMSE=0.7940, MAE=0.5098, MBE=−0.01965). It can be applied to similar hard-rock aquifer regions, particularlyin areas with limited monitoring infrastructure and financial constraints. It provides a foundation for future researchto explore hybrid and advanced machine learning techniques for improved prediction accuracy and model robustness. From a practical perspective, the model serves as an efficient decision-support tool for policymakers, facilitating proactive planning and sustainable utilisation of groundwater resources. The findings emphasise theimportance of intelligent modelling techniques in improving water quality monitoring, safeguarding public health, and ensuring long-term environmental sustainability.
Readers around the world
