Three-vector model predictive power control of doubly fed induction generator based on linear extended state observer under unbalanced grid

Abstract

Doubly-fed induction generator (DFIG) is susceptible to unbalanced grid voltage and mismatched motor parameters during grid-connected operation. The conventional model predictive control (MPC) has low complexity and fast dynamic response, which is widely used in the control of DFIG. However, it has a high steady-state ripple, large computation, and poor robustness. This paper proposes a three-vector model predictive power control based on linear extended state observer (TVMPPC-LESO) to solve the above problems. The method introduces linear extended state observer (LESO) to estimate the system’s lumped disturbance, which makes the calculation of the rotor reference voltage less dependent on the motor parameters to improve the robustness of the MPC. On this basis, the number of switches is decreased and the steady-state ripple is lowered by applying three voltage vectors in a control period and optimizing the switching sequence acting on the rotor-side converter (RSC). By adding a flexible power compensation value to the original power reference value, the TVMPPC-LESO can be extended to unbalanced grids and improve the grid-connected performance of the DFIG. The simulation and experimental results validate its effectiveness by comparing it with conventional MPC, direct power control with space vector modulation based on extended power theory (EXDPC-SVM), and three-vector-based model predictive power control (TV-MPPC).

Keywords

• Doubly-fed induction generator (DFIG)
• Model predictive power control
• Robustness
• Unbalanced grid