Model Predictive Control Using Indirect Operator Splitting Quadratic Program Method and Its Application in PMSM

Continuous Control Set-Model Predictive Control (CCS-MPC) can effectively improve the performance of Interior Permanent Magnet Synchronous Motor (IPMSM), but cannot meet the real-time motor control requirement. Therefore, an indirect Operator Splitting Quadratic Program (OSQP) method is presented, which can solve CCS-MPC quickly and successfully applied to current control of IPMSM. Compared to the active-set method and the interior-point method, the OSQP method does not need division operation after initial matrix decomposition and supports warm starting, which can effectively reduce the computational complexity. Compared with the direct OSQP method, the indirect method only uses half of the data in the matrix operation and reduces the coefficient matrix dimension, which results in higher computational efficiency. In addition, in order to eliminate the static-error caused by the parameters mismatch and suppress the oscillation introduced by building the incremental state equation, the anti-disturbance coefficient and integral coefficient are introduced to effectively solve above problems. The test results of Processor In Loop (PIL) show that the proposed method can solve CCS-MPC quickly and meet the real-time control requirements, and effectively suppress the static-error and oscillation. The performance of the control system is significantly improved.