AI algorithm-based two-stage optimal design methodology of high-efficiency CLLC resonant converters for the hybrid AC-DC microgrid applications

Thanks to the advantages of high power density and the capacity of bidirectional power transfer, the CLLC resonant converter is widely used in the hybrid ac-dc microgrid as a dc transformer to interlink the ac and dc bus. Since the voltages of ac and dc bus are controlled by the energy management system, the CLLC resonant converter operates under open-loop condition, which means the switching frequency and duty cycle are fixed. As a result, in the hybrid ac-dc microgrid applications, for the CLLC converter, the main concern is not the voltage regulation but the conversion efficiency. This paper focuses on the total power loss optimization and the magnetic design of the CLLC resonant converter based on artificial intelligence (AI) algorithm. In order to optimize the total power loss, an AI algorithm-based two-stage optimal design method is proposed. In the first stage, the total power loss, including the driving loss, turn-off loss, conduction loss of the switches, the power loss of the resonant capacitances, and copper and core loss of the transformer are optimized by the proposed AI algorithm, GA+PSO, and the optimal parameters, including the leakage inductances (L_r1 and L_r2), magnetizing inductance (L_m), and resonant capacitances (C_r1 and C_r2) are derived. In the second stage, the optimal leakage inductances and magnetizing inductance are realized by setting proper distance between the primary winding and the secondary winding (d_w), and the thickness of the air gap (d_a). As for the magnetic design, in this paper, the leakage inductances of a planar transformer are used as the resonant inductances. The equations of d_w and d_a to achieve the optimal leakage inductances and magnetizing inductance are derived. Both the proposed optimal design method and the equations of d_w and d_a are validated by simulations and experiments.