MPSO-based PID control design for power factor correction in an AC-DC boost converter

This paper presents the implementation for the first time of a Multi-Particle Swarm Optimization (MPSO) algorithm in the tuning of a PID controller for Power Factor Correction (PFC), applied to a 100W AC-DC boost converter. MPSO algorithm navigates in a search space where each dimension of the space corresponds to the controller constants (Proportional, Integral, Derivative and the Derivative Filter), prioritizing communication over exploration in the algorithm. The controller parameters are randomly initialized in a reduced sector of the space [ , , , ], to optimize the search for a PID solution. In the first step, the algorithm is validated using a simulation model in Simulink and Matlab. Subsequently, a final implementation using a real converter is implemented with the PID tuned by MPSO, improving the PFC obtained in previous work. Although previous works have used evolutionary algorithms applied to heuristic optimization to tunning PID controllers, the MPSO algorithm is not usually used for this purpose, particularly to tunning a PID controller in a power electronics system. One advantage of MPSO over the PSO classical algorithm is the search at different points if the vectorial field looks for an optimal solution. PSO presents problems such as getting stuck in a locally optimal solution. The PID controller is trained offline, with the advantage of allowing the risk of damage in the Boost converter for transitory response, increasing the performance of the Power Factor Correction in the converter. This research opens the possibility to use the extended version of the PSO bioinspired algorithm to tune offline controllers to improve the power converter's performance, minimizing the risk presented in the real-time tuning process.