A Robustness Enhancement Strategy for Three-Vector Model Predictive Current Control Based on Back-EMF Compensation via Sliding Mode Observer

To address the robustness limitations of three-vector model predictive current control (TV-MPCC) in permanent magnet synchronous motor (PMSM) drives under parameter variations and external disturbances, this paper proposes an improved sliding mode observer (SMO) based on a novel dual power-rate reaching law combined with a hyperbolic tangent function (PTHSMO) for back-EMF estimation and feedforward compensation. The proposed reaching law integrates a terminal attractor term and a nonlinear power-rate term to achieve fast convergence, while the tanh-based switching term continuously approximates the sign function to suppress chattering without requiring a downstream low-pass filter. The estimated back-EMF, which encapsulates the combined effect of parameter mismatch and actual back-EMF, is fed forward into the TV-MPCC prediction model to actively compensate for residual disturbances (denoted PTHESMO). The stability of the observer is verified via the Lyapunov method. Compared with the traditional SMO-based TV-MPCC, the proposed method reduces startup overshoot by approximately 46%, decreases speed recovery time under a 0.3 Nm load disturbance from 46.2 ms to 24.5 ms, and reduces rotor position error from 0.1358 rad to 0.1266 rad, providing an effective solution for high-performance sensorless PMSM drive control.