Modeling and Prediction of Water Quality Using Artificial Intelligence Techniques for Enhanced Monitoring and Management.

Abstract

Artificial intelligence offers a cost-effective solution for water supply and sanitation systems while ensuring compliance with quality standards. This study presents a novel AI-based system for real-time monitoring and prediction of water quality, with a focus on sustainability and environmental protection. Using adaptive regression models, the water quality index (WQI) was predicted, and various classifiers were employed to categorize water quality classification (WQC). The dataset included 10 parameters: pH, EC, Ca²⁺, Mg²⁺, Na⁺, K⁺, CO₃ ^2-, HCO₃ ^-, Cl^-, and SO₄ ^2-, which are essential for building efficient prediction models. Among the models tested, linear regression (LR) performed best for predicting WQI, achieving a mean absolute error (MAE) of 2.18 and a coefficient of determination (R²) of 0.967, whereas the K-nearest neighbors (KNN) and random forest (RF) classifiers achieved classification accuracies of 92.3% and 91.6%, respectively, for WQC. The LR model provided accurate WQI predictions, and KNN excelled in classifying WQC. An ensemble model, that combines regression and classification techniques, achieved a high regression coefficient of 96.7% for predicting both the WQI and WQC. This AI-driven approach enhances water quality prediction and supports better water treatment and management systems. These findings have practical implications for improving water quality monitoring and decision-making in pollution-affected regions such as Tamil Nadu, India.