Robust model predictive control of virtual synchronous generators without weighting factors
Su'e Wang,Yufan Liu,Shuai Zhang,Pengfei Hao,Jun Liu
DOI: https://doi.org/10.1002/cta.4117
IF: 2.378
2024-06-01
International Journal of Circuit Theory and Applications
Abstract:This paper proposes a fuzzy decision‐making model predictive control (MPC) strategy with parameter identification using virtual synchronous generator (VSG) as the control object, which can effectively improve the robustness of MPC and avoids the selection of weighting factors in MPC multi‐objective control. Among them, the parameter identification part of this strategy adopts the method of model reference adaptive system, and a feedback correction link is added on the basis of traditional model reference adaptive system (MRAS) to accelerate the convergence speed of MRAS, and the method of fuzzy decision making is adopted to avoid the selection of weighting factors. Summary Virtual synchronous generators (VSGs) can provide damping and inertia for new power systems are of wide interest. Model predictive control (MPC) has the advantages of simple structure, easy to deal with constraints, and can achieve multi‐objective control. However, the conventional MPC has the problems of relying on the model parameters as well as the difficulty of taking the values of the weighting factors. Based on this, this paper proposes a robust MPC control strategy without weighting factors, which improves the system robustness by identifying the model parameters through model reference adaptive system (MRAS); and introduces fuzzy decision making (FDM) to avoid the selection of weighting factors. First, the mathematical model of VSG is established and the prediction model of VSG multi‐objective control is derived. Second, a detailed analysis is conducted to examine the impact of parameter mismatch on VSG. Third, the expression for online identification of parameters is derived based on Popov's superstability theory, using FDM to optimize multi‐objective control and using decision functions instead of cost functions to avoid the selection of weighting factors. Finally, the effectiveness of the proposed control strategy is verified through the RT‐LAB semi‐physical platform.
engineering, electrical & electronic