A Parameter Robustness and Low-Complexity Three-Dimensional Four-Vector Model Predictive Control for LC-coupling Hybrid Active Power Filter

To take full advantage of LC-coupling active power filter (LC-HAPF), Model predictive control (MPC) has been introduced with better performance and fast dynamic response. However, there are two shortcomings, strong parameter dependence and large steady-state error, obstruct its further development. To solve these problems, this article proposed strong parameter robustness and low-complexity three-dimensional (3D) four-vector model predictive control (FV-MPC) for LC-HAPF in three-phase four-wire systems. First, an ultralocal model of LC-HAPF is established with a simple nonlinear disturbance observer (NDO) to estimate and compensate for the lumped disturbance, which eliminates additional ac capacitor sensors and enhances parameter robustness without heavy history data or complex parameters design. Then, 3D FV-MPC method is further proposed, with the support of equivalent transformation between the switching function and duty cycle, the groups of voltage vectors and computational complexity can be significantly reduced. Besides, the optimal duty cycle calculation with the derivation method can ensure better performance. Finally, experimental results demonstrate that the proposed approach enhances transient and steady-state performance while reducing CPU burden by 30%, simultaneously, remains low current error within ±60% inductance perturbation, and makes it ideal for higher-level systems.