Prediction of Concentration of Iron Using Linear-Nonlinear Hybrid Models

Accurate predictions of groundwater iron concentration are essential for mitigating health risks, ensuring effective water treatment, minimizing environmental impact, enabling sustainable resource management, and maintaining infrastructure integrity. In this study, Multi-Layer Perceptron Neural Networks (MLPNN) and Multiple Linear Regression (MLR) are used for predicting the concentration of iron (Fe) in groundwater. Besides, a hybrid model combining MLR and MLPNN is proposed to predict Fe to enhance the accuracy of individual models. To this aim, a total of 192 representative groundwater samples were collected from five separate investigations specifically conducted to assess groundwater quality. Based on the results, it has been observed that the proposed linear-nonlinear hybrid models demonstrate superior statistical performance in terms of predicting accuracy compared to all models considered. The proposed model shows potential for application in similar studies focused on predicting groundwater quality.