Hierarchical learning-based swarm optimization for control system design of three-phase rectifier circuits

Nowadays, evolutionary algorithms have been widely applied to design and estimate the parameters of power electronic systems. However, for closed-loop control systems, selecting and balancing the weights of performance indicators remains challenging for evaluating the whole performance of three-phase rectifiers. Additionally, classical evolutionary algorithms are prone to local optima. To address these challenges, a hierarchical learning-based swarm optimization algorithm (HLSO) is proposed to obtain optimal PI parameters, thereby improving the whole performance of three-phase rectifiers. First, the fitness function incorporates performance indicators such as settling time, steady-state error, and overshoot to comprehensively evaluate the system's control performance. Then, the hierarchical learning and exemplar selection strategy enhances the exploration capability to avoid local optima. After that, the particle update rule is established, where two particles from different higher levels are randomly selected to guide the learning process of other particles and avoid premature convergence. Finally, the effectiveness of the proposed method is verified by simulation.