Utilizing Neural Network for Designing Load‐Independent Class E Inverter

This paper presents the design, simulation, and fabrication of a load‐independent Class E inverter utilizing an artificial neural network (ANN) facilitator. The load‐independent inverter is intended for wireless power transfer systems, such as wireless charging for small electronic devices, medical applications, robotics and electric vehicles. Class E inverters are favoured due to their high efficiency, low‐cost design and compatibility with resonant power transfer circuits; however, they face challenges such as sensitivity to load changes and complexity in theoretical calculations. Also, the design process addresses the presence of parasitic nonlinear elements and harmonic distortions, making it time‐consuming and potentially less accurate. To overcome these challenges, a two‐layer ANN is employed to determine the structural parameters of the circuit, resolving issues in the design process and ensuring sufficient accuracy despite parasitic components and maintaining a constant output voltage regardless of load resistance variations. The ANN accelerates the design process by calculating inverter parameters based on the desired operating frequency. Experimental results show strong agreement with theoretical and simulation outcomes, validating the proposed inverter design. The proposed load‐independent Class E inverter has an output power of 2 W at a load resistance of 15 Ω and a DC input voltage of 5 V at a frequency of 1 MHz.