A novel robust dead-beat structure for double vector model predictive control in three-level inverter fed PMSM drives

Abstract

Model predictive control (MPC) has recently been considered in permanent magnet synchronous motor (PMSM) drives due to its rapid dynamics and simple structure. However, conventional MPC utilizes only a single voltage vector per control cycle, and the prediction model relies heavily on the motor parameters. Hence, substantial current ripples and poor disturbance rejection have largely limited its adaptability to a variety of environment conditions. This article presents a novel double vector MPC (DV-MPC) scheme for three-level inverter fed PMSM drives. To reduce the computational complexity of DV-MPC, the cost function is derived with the reference voltage by employing the dead-beat approach, and the inverter's neutral point potential is balanced using the complementary small voltage vectors. The dead-beat voltage prediction model is further enhanced by incorporating an active damping framework and an extra coefficient to enhance the robustness against parameter variations and disturbance rejection. Moreover, a robust current predictor is designed for the delay compensation. The proposed method is straightforward to implement and achieves strong disturbance rejection and parameter robustness with a fast dynamic response. Experimental results demonstrate the effectiveness of the proposed method.