A Machine-Learning-Assisted Localization and Magnetic Field Forming for Wireless Powering of Biomedical Implant Devices

Abstract

This article presents an array of overlapped coil transmitter antenna for localizing and generating desired magnetic beams toward the localized receiver to address misalignment problems in near-field wireless power transfer (WPT) applications. For this purpose, a time-divisional approach is employed to obtain the voltage samples from the moving receiver for predicting the position and orientation of the receiver using a machine-learning algorithm. In contrast, particle swarm optimization (PSO) is utilized to obtain the optimal current distribution to construct a magnetic beam in the receiver direction. The proposed transmitter is also optimized to attain a highly nonuniform magnetic field distribution to improve the localization sensitivity and generate a sharp magnetic beam toward the receiver. The proposed transmitter is fabricated using a high-frequency litz wire and excited using a single source of excitation and switching circuitry. The switching circuitry enables the extraction of voltage samples for localization purposes and constructs the desired magnetic beam. The performance of the fabricated prototype is measured experimentally, which corroborates with analytical results. The results demonstrate the potential of the proposed transmitter to achieve a misalignment-resilient WPT for charging small devices compatible with biomedical implants, wireless endoscopy capsules, and wearable devices.