Model predictive control of matrix converter based solid state transformer

A model predictive control (MPC) is proposed for two three-to-single-phase matrix converters (MC) in back-to-back connection based solid state transformer (SST). The MC-SST system presents single-stage ac-ac conversion, reduced transformer volume and weight compared to using the conventional line-frequency transformer, and smart transmission capability due to the controllable voltage amplitude and frequency. However, the existing modulation methods contain sophisticated voltage and current vectors computation and duty cycle composition, which could make the control of the proposed topology too complex. In this paper, it is decided to use easy to perform MPC for the whole scheme. The proposed MPC predicts the circuit variables based on the discrete model of MC-SST system, and evaluates a cost function that ensures closest value to the current reference of ac outputs. The best switching state is selected to drive the power devices in the next sampling period. Simulation studies carried out on different voltage amplitude and frequency between ac inputs and outputs, as well as load power variations demonstrated the effectiveness and simplicity of the proposed MPC for the MC-SST.