Predicting the environmental risks of potentially toxic metal(loid)s in mechanochemically treated fly ash using machine learning.

Mechanochemical (MC) treatment is a green and efficient approach that has shown great potential in reducing the environmental risks of potentially toxic metal(loid)s (PTMs) in municipal solid waste incineration fly ash (MSWIFA). However, accurately predicting the environmental risks of PTMs and optimizing the operating conditions remain challenge. In this study, six machine learning (ML) models were employed to predict the environmental risks of PTMs in MC-treated fly ash, quantified by the Overall Pollution Toxicity Index (OPTI). In particular, the eXtreme Gradient Boosting (XGB) model achieved the best performance, with R2 values of 0.986 for the training set and 0.921 for the test set, indicating excellent predictive accuracy and generalization. Feature importance analysis revealed the following ranking of influence on environmental risks: additives (48.2 %) > MC conditions (28.7 %) > PTMs properties (21.3 %) > fly ash composition (1.8 %). MC treatment time, initial concentration, Ca-based additive, P-based additive, Ca-P-based additive, Si-Al-based additive, leaching pH, and Cl were the eight important input features for predicting environmental risks of PTMs in MC-treated fly ash. Furthermore, a graphical user interface (GUI) was developed based on the trained models, enabling rapid environmental risks assessment and process optimization. This study presented a data-driven framework for predicting environmental risks in MC-treated fly ash and provided practical tools to support engineering-scale applications. The development of green and efficient additives represents the most effective approach for addressing the PTMs issues in MSWIFA.