In recent years, wireless power transmission systems that do not require charging cables have become widely used, especially in electric vehicles, smartphones, and medical devices. To achieve efficient design automation, an AI-based prediction system has been proposed for the design of transmitters (receivers) used in wireless power transmission systems. However, conventional AI models are limited to predicting transmission efficiency and cannot predict multiple electrical characteristics of the system (e.g., operating frequency, coupling performance). Furthermore, the system was limited in size and transmission distance, thus limiting its application.
A research team led by Professor Pokharel Ramesh, Assistant Professor Adel Barakat, and third-year Doctoral Student Xin Jiang in Kyushu University's Graduate School of Information Science and Electrical Engineering, has used AI to predict the frequency and transmission distance dependent electrical characteristics of wireless power transmission systems for the first time, thereby shortening the time for system design. The results were published in IEEE Transactions on Antennas and Propagation.
Based on the concept of AI translation, the newly developed system "translates" the layout, electrical elements, and transmission distance of the transmitter and receiver into the "form" of electrical characteristics (S-parameters) that vary with the frequency and electrical elements. The AI model proposed here can predict and apply more electrical characteristics than conventional models, enabling system design automation. Efficient design automation allows the user to simply enter the required electrical parameters and the system layout is automatically generated based on them. Specifically, in the case of wireless power transmission systems, the AI system considers the parameters and generates an optimal layout when given a specific transmission distance and operating frequency. This will greatly reduce the time and effort required for design work and improve the efficiency of the design process.
Journal Information
Publication: IEEE Transactions on Antennas and Propagation
Title: An Efficient Inverse Modeling Method Using Translator-Inspired Neural Network and Dual-Annealing for a Compact WPT System
DOI: 10.1109/TAP.2024.3372149
This article has been translated by JST with permission from The Science News Ltd. (https://sci-news.co.jp/). Unauthorized reproduction of the article and photographs is prohibited.