Bat Algorithm–Based Optimization of Deep Models for Heavy Metal Detection in Wastewater

Abstract

Heavy metal contamination in wastewater poses significant environmental and health risks, necessitating accurate and efficient detection methods. This study presents a novel approach combining Bat Algorithm (BA) optimization with deep learning models for predicting heavy metal concentrations in industrial wastewater. The proposed BA-optimized Long Short-Term Memory (LSTM) network demonstrates superior performance in detecting six heavy metals (Cu, Zn, Pb, Cd, Cr, Ni) compared to conventional machine learning approaches. Real datasets from industrial wastewater treatment plants were analyzed, comprising 1,250 samples collected over 18 months. The BA optimization algorithm successfully tuned Hyperparameters of the deep learning model, achieving an R² of 0.968 and RMSE of 0.142 mg/L. The results indicate that the proposed hybrid model outperforms traditional methods with R² improvements of 0.12-0.18 while reducing computational time by 35%. This research contributes to the development of intelligent monitoring systems for wastewater treatment plants, enabling real-time heavy metal detection and proactive environmental management.