Artificial intelligence and machine learning-based modeling and control of wireless power transfer systems

Abstract

Wireless power transfer (WPT) technology has emerged as a promising solution for numerous applications, including electric vehicles, consumer electronics, and medical devices. This paper focuses on the application of advanced artificial intelligence (AI) methods, specifically Random Forest and XGBoost, to predict key parameters in the Series Series (SS) topology-based wireless charging system. Experimental data collected from both a physical model of WPT platform and PSIM simulation platform are utilized to enhance the accuracy and reliability of the models. By training on a dataset split into 70% training data and 30% testing data, the AI models learn the relationships between the voltage and current harmonic components of the transmitting and receiving coils. The trained models effectively predict the mutual inductance and load resistance values, which are critical for system optimization. The results demonstrate that these AI methods significantly reduce the influence of noise and enable accurate measurements at resonance frequencies, addressing challenges in traditional measurement techniques. Metrics such as Mean Squared Error (MSE) and Root Mean Squared Error (RMSE) validate the high accuracy and efficiency of the regression models. The comprehensive dataset obtained through both simulation and physical experimentation ensures robust model performance and highlights the practical applicability of the proposed approach. The findings underline the benefits of leveraging AI in modern power electronics systems, paving the way for more efficient and reliable wireless charging solutions.