Modeling textile dye removal by ultrasound‐assisted fenton process using machine learning approaches

Abstract Textile wastewater contains harmful synthetic dyes, organic contaminants, and toxic chemicals, requiring advanced treatment methods to protect the environment and human health. It is important to find the best way to use new treatment methods for textile wastewater. In order to forecast the textile dye removal effectiveness utilizing the ultrasound‐assisted Fenton procedure, this study used machine learning (ML). A test set was used to assess the model's prediction skills after it was developed using a train set. Subsequently, a prediction model was constructed to understand the effect of each feature on remediation efficiency. According to the laboratory experimental results obtained, it was determined that pH 2 was the most suitable value for the most effective removal of organic contaminants in the Fenton process, and it was revealed that pH was the factor with the greatest impact on removal efficiency. The created Linear Regression (LR) model showed promise in forecasting textile dye removal efficiency with the use of the ultrasound‐assisted Fenton procedure. The LR model achieved an accuracy score of 0.9265 for removal efficiency, with a Mean Squared Error (MSE) of 90.664, a Mean Absolute Error (MAE) of 7.489, and a Root Mean Squared Error (RMSE) of 9.522. In general, this research would aid in assessing textile dye's remediation capacity and maximize removal efficiency under ideal circumstances.