Robust Adaptive RBF Neural Network Speed Control of PMSM Integrating Hybrid Nonlinear Reaching and Composite Observation

Complex industrial uncertainties significantly impair fast-dynamic speed regulation in electric drives. To address these challenges, this paper proposes a unified adaptive sliding-mode control framework for permanent magnet synchronous motor (PMSM) speed regulation. Specifically, an advanced hybrid nonlinear reaching law (HNRL) is developed to enhance transient response while effectively suppressing chattering, thereby guaranteeing uniform practical fixed-time reaching. Furthermore, a hybrid nonlinear disturbance observer (HNDO) is designed to accurately estimate slow-varying disturbances, while a hybrid nonlinear radial basis function neural network (HNRBFNN) is introduced to compensate for fast-varying disturbances. Distinct from existing RBFNN-assisted controllers, the proposed HN-RBFNN reuses the intrinsic nonlinear term of the HNRL as the driving signal for both the network input and the adaptive update law. This unified design enables fixed-time convergence to a shrinking neighborhood, whose ultimate bound automatically decreases over time owing to the exponential decay of the HNDO estimation error. Experimental results demonstrate the effectiveness of the proposed method in tracking performance, disturbance rejection, harmonic suppression, and composite observation capability.