AI driven adaptive resonance compensation for high efficiency wireless power transfer in electric vehicle charging

Wireless Power Transfer (WPT) has emerged as a transformative charging technology for electric vehicles (EVs) due to its convenience, safety, and operational efficiency. However, existing WPT architectures often suffer from performance degradation caused by coil misalignment, resonance detuning, load fluctuations, and external electromagnetic interference, which limit their adaptability and reliability in dynamic charging environments. To address these challenges, this study proposes a hybrid Feedforward Neural Network (FNN) and Temporal Convolutional Network (TCN) framework optimized using Particle Swarm Optimization (PSO) to enhance WPT system performance. The FNN effectively processes static system parameters such as coil geometry, compensation topology, and operating frequency, while the TCN captures temporal dependencies associated with dynamic alignment variations and load changes. PSO is employed to optimize network hyperparameters, ensuring improved predictive accuracy and computational efficiency. The proposed framework is validated using a real-world EV charging dataset augmented with synthetic perturbations to emulate WPT-specific operating conditions. Experimental results demonstrate superior performance, achieving 96.8% power transfer efficiency, an R² value of 0.99, and minimal prediction errors with MAE of 0.012 and RMSE of 0.024. Furthermore, the system maintains low inference latency of 15 ms and total computation time of 120 s, supporting real-time deployment. Implemented in Python with robust simulation libraries, the proposed model offers a scalable and reliable solution for next-generation intelligent WPT-enabled EV charging systems.