AI-Driven prediction of sulfide removal from wastewater through aeration and iron dosing

Abstract

Wastewater containing sulfide needs to be segregated to save the environment and preserve the quality of water. Precise prediction of sulfide is important to monitor the environment, the exploration of minerals, and industrial operations. In this study, five widely used machine learning (ML) models namely, XGBoost, Artificial Neural Networks (ANN), Support Vector Regression (SVR), Random Forest (RF), and AdaBoost, using the published data of suldife removal by aeration and iron salts. XGBoost was selected as the best model (RMSE: 160.86, R²: 0.973) and was much better than others due to hyper parameter optimization and extensive analysis (RMSE, MAE, R²). The correlation study of features indicated that there are fundamental input relationships that dictate the behavior of sulfide. The optimized XGBoost structure achieves highest predictive accuracy. The results explain the potential application of machine learning models to optimize waste water treatment parameters and increase efficiency.