Adaptive Moment Estimation-Based Model Predictive Control for PMSM Motor With Low Tracking Error

To address the tracking error issue in finite control set model predictive control (FCS-MPC) systems, this article proposes an online optimization MPC method for permanent magnet synchronous motor (PMSM) based on adaptive moment estimation (Adam). A critical innovation is the integration of the Adam algorithm into the real-time solution of MPC, adaptively tailored to PMSM control characteristics. Specifically, the proposed method designs an MPC cost function in a convex quadratic programming (QP) form, incorporating PMSM control constraints. It leverages the momentum and adaptive learning rate mechanisms of the Adam optimizer to solve the optimization problem online efficiently. Simultaneously, a momentum prediction mechanism is further introduced into the bias correction part of moment estimation, improving estimation accuracy and convergence speed. Furthermore, a dynamically decaying first-order moment coefficient is also employed to ensure the real-time convergence of the proposed algorithm. Experimental results demonstrate that the proposed method effectively reduces current tracking error, improves current waveform quality, and exhibits excellent real-time control potential.