Range-adaptive Impedance Matching of Wireless Power Transfer System Using a Machine Learning Strategy Based on Neural Networks

This work describes the implementation of a machine learning (ML) strategy based on the neural network for real-time range-adaptive automatic impedance matching of Wireless Power Transfer (WPT) applications. This approach for the effective prediction of the optimal parameters of the tunable matching network and classification range-adaptive transmitter coils (Tx) is introduced in this paper aiming to achieve an effective automatic impedance matching over a wide range of relative distances. We propose a WPT system consisting of a tunable matching circuit and 3 Tx coils which have different radius controlled by trained neural network models. The feedforward neural network algorithm was trained using 220 data and classifier's in pattern recognition accuracy were characterized. The proposed approach achieves a Power transfer efficiency (PTE) around 90% for ranges within 10 to 25cm, is reported.

I. Introduction

The impedance matching of a wireless power transfer (WPT) system using magnetic resonance coupling (MRC) has become a critical challenge in order to maintain a reasonable Power transfer efficiency (PTE) for time-varying configurations. Several approaches of impedance matching have been proposed [1], [2] and [3] regarding the distance between the receiver (Rx) and transmitter (Tx) as PTE varies significantly with distance. However, these are limited in the effective ranges as a consequence of their unexpected variation of the transfer distance or load impedance. Here, we propose an alternative approach that takes advantage of a novel method based on a feedforward neural network combined with pattern recognition techniques, thus addressing the shortcomings of the aforementioned impedance matching approaches while retaining high PTE. As a proof-of-concept, one receiver coil, three selective transmitter coils and a matching circuit with tunable capacitors are first designed and measured. Then, a machine learning approach utilizing neural network algorithms that can construct the mapping relationship is presented to improve the capability of the WPT system.