Intelligent Edge AI Framework for Foreign Object Detection and Power Optimization in WPT

The wireless power transfer (WPT) invention has received considerable attention because of its practical applications and promising potential in electric vehicles and consumer electronics. However, safety and efficiency remain major concerns, mainly because of the risk of foreign object interference and uncontrolled power supply. This paper proposes an intelligent WPT system that integrates artificial intelligence (AI)-based Foreign Object Detection (FOD) and Dynamic Power Control (DPC) to ensure safe and efficient operation. The system uses a lightweight convolutional neural network (CNN), implemented on an edge computing platform, to detect and classify metallic or dangerous foreign objects in real time using data from thermal and vision sensors. It dynamically adjusts or stops power transfer upon detection to prevent overheating or loss. In addition, the system uses a self-adaptive power control algorithm that learns load patterns and environmental information to optimize power conditions based on coil alignment, load demand, and safety constraints. Experimental results demonstrate improved sensing accuracy, reduced power loss, and enhanced thermal protection compared to traditional distance-based methods. This approach provides a scalable, low-latency solution for next-generation WPT systems in automotive, consumer, and industrial applications.