Modeling of groundwater quality index by using artificial intelligence algorithms in northern Khartoum State, Sudan

Abstract In the present study, multilayer perceptron (MLP) neural network and support vector regression (SVR) models were developed to assess suitability of groundwater for drinking purposes in northern Khartoum area, Sudan. The quality was evaluated by predicting index (GWQI). GWQI is a statistical model that uses sub-indices accumulation functions reduce dimensionality data. first stage, calculated using 11 physiochemical parameters collected from 20 wells. These include pH, EC, TDS, TH, Cl ? , SO 4 ?2 NO 3 Ca +2 Mg Na + HCO . primary investigation confirmed all except EC are beyond standard limits World Health Organization (WHO). measured ranged 21 396. As result, samples classified into three classes. majority samples, roughly 75%, projected excellent water category; 20% considered good 5% as unsuitable. powerful tools assessment; however, computation lengthy, time-consuming, often associated with calculation errors. To overcome these limitations, this study applied artificial intelligence (AI) techniques develop reliable prediction employing MLP SVR models. input data detected parameters, output computed GWQI. dataset divided two groups ratio 80% training validation. predicted actual (calculated GWQI) compared four criteria, namely, mean square error (MSE), root squared (RMSE), absolute (MAE), coefficient determination ( R 2 ). Based on obtained values performance measures, results revealed robustness efficiency modeling Consequently, north area suitable human consumption BH 18, where highly mineralized observed. approach advantageous evaluation recommended be incorporated modeling.