Development of hybrid ANFIS-GAN-XGBOOST models for accurate prediction of material removal rates from PCB-polluted concrete surfaces using laser technology for sustainable energy generation

Abstract

This study presents a hybrid machine learning approach for predicting material removal rates from polychlorinated biphenyl (PCB)-polluted concrete surfaces using laser technology. An adaptive neural fuzzy inference system (ANFIS) with XGBoost and generative adversarial networks (GANs) was employed as a hybrid model (ANFISGAN- XGBoost) to classify the numerous input variables for material removal rates based on process factors. Four input parameters, including annealing temperature, scanning speed, frequency, and laser power, were evaluated to calculate the ablation effectiveness and material removal rates. The study demonstrated that the hybrid ANFIS-GAN-XGBoost model applied to laser ablation on concrete surfaces polluted with PCBs accurately predicts the ablation depth. The experimental findings revealed that 96.66% of the PCBs were removed from the concrete surface, and 88.43% of them were decomposed during the laser decontamination procedure. A surface ablation ratio of roughly 7.2 m2/h was achieved, indicating the potential of using laser technology for sustainable energy generation by removing pollutants from concrete surfaces. The findings of this study contribute to the development of sustainable energy generation strategies and could be used in the cleanup of concrete surfaces polluted with PCBs in various settings, including nuclear power plants and polluted buildings. The hybrid ANFIS-GAN-XGBoost model is a valuable tool for predicting material removal rates based on process factors and can be used in the optimization of the laser-based decontamination process for PCB-polluted concrete surfaces. This study highlights the potential of using advanced machine learning models in environmental cleanup efforts, paving the way for future research in this field.