Weighting factor design based on SVR-MOPSO for finite set MPC operated power electronic converters

• Published in: Journal of power electronics Vol. 22; no. 7; pp. 1085 - 1099
• Main Authors: Liu, Yonglu, Yang, Zhengmao, Liu, Xubin, Dan, Hanbing, Xiong, Wenjing, Ling, Tao, Su, Mei
• Format: Journal Article
• Language: Korean
• Published: 2022
• Subjects: Multi-objective particle swarm optimization (MOPSO); Support vector regression (SVR); Weighing factors design; Finite set model predictive control (FS-MPC); Voltage source converter (VSC)
• ISSN: 1598-2092; 2093-4718

Selecting weighting factors is a challenge for the finite set model predictive control (FS-MPC). Based on the support vector regression (SVR) algorithm and the multi-objective particle swarm optimization (MOPSO) algorithm, this paper proposes a new weighting factor design principle. SVR is used to establish the functional relationship between the input weighting factors and the output performance indexes (such as the average switching frequency (fsw) and the total harmonic distortion of the output voltage). Even in the case of small samples, this can provide accurate performance index estimates for any combination of weighting factors. The established SVR function is taken as the fitness function. Then, MOPSO is used to search for Pareto optimal weighting factor combinations. The proposed method can converge in a few steps and does not require tedious calculations. Moreover, it is applicable to optimization problem with two or more weighting factors for arbitrary topology models. It also provides a range of optimal weighting factor solution sets. Finally, the proposed methodology is verified on a practical weighting factor design problem in a FS-MPC regulated voltage source inverter. Experimental results confirm the correctness of the theoretical analysis.