Comparative Analysis of Machine Learning Models and Explainable Artificial Intelligence for Predicting Wastewater Treatment Plant Variables

Advances in Environmental and Engineering Research, Год журнала: 2024, Номер 05(04), С. 1 - 23

Опубликована: Окт. 17, 2024

Increasing urban wastewater and rigorous discharge regulations pose significant challenges for treatment plants (WWTP) to meet regulatory compliance while minimizing operational costs. This study explores the application of several machine learning (ML) models specifically, Artificial Neural Networks (ANN), Gradient Boosting Machines (GBM), Random Forests (RF), eXtreme (XGBoost), hybrid RF-GBM in predicting important WWTP variables such as Biochemical Oxygen Demand (BOD), Total Suspended Solids (TSS), Ammonia (NH₃), Phosphorus (P). Several feature selection (FS) methods were employed identify most influential variables. To enhance ML models’ interpretability understand impact on prediction, two widely used explainable artificial intelligence (XAI) methods-Local Interpretable Model-Agnostic Explanations (LIME) SHapley Additive exPlanations (SHAP) investigated study. Results derived from FS XAI compared explore their reliability. The model performance results revealed that ANN, GBM, XGBoost, have great potential variable prediction with low error rates strong correlation coefficients R² value 1 training set 0.98 test set. also common each model’s prediction. is a novel attempt get an overview both LIME SHAP explanations

Язык: Английский