Capacitance Prediction Using Multi-cascade Convolutional Neural Network for Efficient Wireless Power Transfer

Abstract

The efficiency of the wireless power transfer is significantly impacted by misalignment between the transmitting and receiving coils due to impedance mismatching. To tackle this issue, an efficient power transfer solution is proposed, employing a capacitance prediction method based on a multi-cascade convolutional neural network. In the study, the impedance matching characteristic of a magnetic coupling resonant wireless power transfer system with an impedance matching network is analyzed. After that, a neural network-driven approach is introduced to establish a mapping between reflection impedance and the optimal capacitance, and the impedance matching performance of the system is assessed in the presence of coil misalignments.