Abstract:

As the existing digital control methods for active power filters are highly dependent on the adoptea models, this paper proposes a model predictive control （MPC） method for a three-phase four-leg active power filter based on the decoupling control of the positive-negative-order and the zero-order systems. By taking the difference equation as the predictive model, the feedback correction and dynamic optimization in the conventional MPC methods are reserved, and the optimal control law is deduced. The proposed method is then compared with common predictive current control methods, drawing a conclusion that the predictive current control with supply voltage observer is only a special case of MPC. Moreover, the influences of system parameters on the system stability are analyzed, and a larger modeling error allowance of connection reactor and a stronger robustness of the controller are ob- tained by modifying MPC parameters, with the bandpass filter being omitted. The combination of MPC and a repetitive controller eliminates the inherent steady-state error of MPC. Simulated results indicate that the proposed method is effective and valid.

Key words: three-phase four-leg power, active power filter, model predictive control, repetitive control, robustness