Using artificial neural network to predict degradation rates of pollutants in industrial wastewater with TiO2-based nanocomposites

Abstract Wastewater treatment, especially for industrial wastewater, has always been a topic of interest due to water resource limitations and high levels of wastewater production. Using photocatalytic materials like TiO 2 is an economical and efficient method. Additionally, doping TiO 2 with metallic and non-metallic elements can extend its absorption spectrum from ultraviolet light to visible light, allowing for better utilization of natural sunlight. However, experimental methods for testing these processes can be costly and challenging to implement. To address this, we have developed a multi-layer perceptron artificial neural network model to predict the degradation rate of industrial wastewater pollutants. Nine key input parameters were used to train Degradation Predictor Neural Net (DPNN) model, and the results show that the DPNN achieved the highest prediction accuracy, with MSE = 0.085, MAE = 0.058, RMSE = 0.092, and R 2 = 0.943, significantly outperforming the other models. These findings confirm that the DPNN provides a reliable and robust framework for predicting photocatalytic degradation processes and can be extended to other pollutants beyond those studied here.