Volume 41: Energy Transitions toward Carbon Neutrality: Part IV

Predicting the Degradation Kinetic Constants of Organic Pollutants in Sonoelectrochemical System using Machine Learning Methods Yongyue Zhou, Yangmin Ren, Shiyu Sun, Fengshi Guo, Mingcan Cui, Jeehyeong Khim



The petroleum industry is one of the fastest-growing sectors and has made a significant contribution to the economic growth of developing countries. Wastewater generated by the petroleum industry contains a variety of organic pollutants. These organic compounds exist in highly complex forms in discharged water and cause environmental hazards. Sono-electrochemical system is emerging as a future trend due to its clean and non-secondary pollution characteristics. This process combines ultrasound and electrochemical methods to enhance the reaction rate constants for pollutant degradation. However, in the system design process, the complex interactions between EC, US, pollutants, and environmental parameters significantly impact the outcomes. Therefore, predicting the kinetic constants of organic compound degradation in US-EC systems within complex reaction systems is challenging. In this study, Machine learning models such as Artificial Neural Networks (ANN), Support Vector Machines (SVM), and Extreme Gradient Boosting (XGBoost) were employed to predict the degradation rates of organic compounds in US-EC systems. Comparative analysis of the prediction results from different models showed that XGBoost performed exceptionally well, with R2 and RMSE values of 0.97 and 0.0006, respectively. SHAP analysis was conducted to evaluate the impact of design parameters on the model’s predictive performance, and the results indicated that ultrasonic frequency, ultrasonic power, and the distance ‘r’ from the ultrasonic transducer to the electrode had the most significant influence on the model’s predictive performance. This method effectively guides the parameter design of US-EC systems and enables accurate predictions of the degradation rates of organic compounds.

Keywords Sonoelectrochemical, Design parameter, Machine learning, SHAP model, Organics degradation, Prediction method

Copyright ©
Energy Proceedings