A Robust Model Predictive Current Control Using Cascaded Structure of Noise-Resilient Augmented Observer With Adaptive Data-Driven Iterative Harmonic Identification for PMSM Drives

A robust model predictive current control (MPCC) strategy is proposed for motor drive systems, in which a cascaded architecture is innovatively constructed by integrating a noise-resilient augmented observer (NAO) with an adaptive data-driven iterative harmonic identification (ADIHI) mechanism. First, to address dominant aperiodic disturbances in the ultralocal motor model, an enhanced NAO is developed, where the integral of noisy measured current is deliberately introduced as an augmented state variable. This novel augmented state effectively decouples the adverse interaction between high observer gains and high-frequency current measurement noise, enabling inherent low-pass filtering properties of the integrator operator and significantly improving noise immunity without sacrificing the disturbance estimation accuracy. Second, based on the NAO-estimated disturbances and ultralocal model, a high-pass-filtered harmonic internal model component is extracted and further refined through the proposed ADIHI scheme. By incorporating an adaptive forgetting factor and an iterative identification mechanism, ADIHI achieves accurate online harmonics extraction. Moreover, an adaptive forgetting factor updating strategy driven by the variation rate of harmonic identification error is designed to enhance robustness against operating condition variations. Finally, the overall closed-loop system stability of the proposed MPCC framework is rigorously established using the Nyquist stability criterion. Explicit analytical expressions for control parameters are derived under stability constraints, and systematic stability regions and parameters tuning guidelines are provided. Extensive experimental results on a motor test bench demonstrate the effectiveness and superiority of the proposed current control strategy.