Japanese scientists have developed a new, highly efficient method for designing wireless power transfer (WPT) systems. Based on machine learning, this method allows the system to maintain a stable voltage even when the load changes. Wireless Power Supply.
WPT systems are already an important part of many devices, from smartphones and biomedical sensors to inductive Stovetops, from using WPT mechanisms to heat cookware. But one of the key issues with technology today is the struggle to fluctuate power. This is because they are load dependent. In other words, system performance is greatly influenced by which devices are installed (load).
Devices like smartphones rely on a certain regulatory voltage to safely charge the battery. The battery's resistance to power changes upon filling. This can cause voltage fluctuations in load-dependent WPT systems. This can damage the device or reduce the charging speed.
In comparison, the new machine learning approach is load-independent (LI), meaning that it can provide consistent power and maintain high efficiency no matter which device is charged. In the example of a smartphone battery, this means that power continues to be efficiently transferred at a stable voltage regardless of the resistance fluctuations that can occur when the battery is full.
This is especially important for larger batteries in more complex applications, such as electric vehicles, where loads can change dramatically during charging.
Researchers revealed their findings in a new study published in the journal in June 2025 IEEE Transactions on Circuits and Systems.
Related: DARPA destroys wireless power records and paints energy over 5 miles away – use it to make popcorn
A WPT system works through a process called resonance, just as radio and television receive over-the-air signals. The transmitter broadcasts certain frequency of electromagnetic waves by adjusting the way the power vibrates between the capacitor and the inductor. When the wave reaches a receiver circuit set to the same frequency, the two resonate and amplify the signal significantly.
Radios send signals to other components to further amplify and demodulate the sound, while in WPT systems, this resonance allows receivers to capture and store the energy transmitted wirelessly.
Machine learning to increase wireless power
The new approach uses machine learning to model and optimize less load-dependent power transfer systems. This process involves artificial intelligence observing while building a virtual model of the system and performing simulations of the model.
AI determines how well a system is operating based on criteria such as the amount of electricity as heat and how clean the electrical signals are. Next, using trial and error methods, the system is optimized to operate at peak efficiency, and transfer power with minimal variation and energy dissipation.
Researchers used a new method to reduce the variability to 5% compared to 18% using a load-dependent system. Additionally, the power transfer efficiency has been increased to 86.7%, but load dependent systems work 65% less efficient.
Load-independent WPT systems have broader meanings than wireless charging devices, the research lead author said Hiroo sekiyaProfessor at the Graduate School of Advanced Integrated Science at Chiba University.
“We believe the results of this study are an important step towards a fully wireless society,” he said. statement. “In addition, the behavior of LI allows WPT systems to be built in a simple way, reducing costs and size. Our goal is to make WPT general within the next five to ten years.”
This study shows how AI can be used to improve the design of electrical circuits, leading to the transformation of “power electronics design” that moves towards the future of automated circuit design.
