Foreign Object Detection for Wireless Power Transfer Based on Machine Learning

A simple foreign object detection (FOD) method utilizing a neural network is discussed. For a receiver and transmitter structure, open helical types of coil resonating at 13.56 MHz were used with different distances and with and without the presence of foreign objects (a copper plate or plastic bottle filled with water). An FOD system was constructed based on a neural network that detects foreign objects based only on reflection coefficient (S11) parameters. This approach for FOD achieved an accuracy of approximately 98% in the range 12-18 cm.