An Improved Flyback Resonant DC–DC Converter and Its Multiobjective Parameter Optimization Method

Flyback converter is widely utilized for low-power dc–dc conversion with the simple circuit structure. However, it suffers severe switching loss caused by hard switching operation with the enhanced switching frequency. This article proposes a 10 MHz resonant dc–dc converter by ameliorating the traditional flyback converter. The converter employs a resonant inductor and two resonant capacitors, which achieves soft switching of both the switch and the diode. Furthermore, the circuit performance gets further improved by reversing the connection of the transformer dotted terminals. To solve the problem that the parameter design becomes thorny due to the introduction of multiple resonant components, a multiobjective parameter optimization design method is proposed, which combines back prorogation neural network (BPNN) and genetic algorithm (GA). It simplifies the design process and improves the design accuracy. Finally, two 10 MHz prototypes with 12 V input and 5 V/10 W output were built to verify the theoretical analysis.