Robust predictive torque control of switched reluctance motor based on linear extended state observer

To enhance the performance and robustness of predictive torque control amidst modeling errors and parameter variations in switched reluctance motor (SRM) drive systems, this paper proposes a model-free predictive torque control strategy utilizing a linear extended state observer. Initially, a novel torque error dynamic compensation method is introduced, enabling accurate mapping of phase torque to phase current. This method is characterized by its simplicity, ease of parameter setting, and its capability to bypass the complexities of solving the torque inverse model. Subsequently, an improved model-free predictive control algorithm is developed for current regulation. This algorithm substitutes the SRM’s nonlinear model with a super local model and employs a linear extended state observer to estimate internal disturbances, such as model errors and parameter variations. The primary advantage of this algorithm is its data-driven nature, eliminating the dependence on precise mathematical models of the motor drive system. Ultimately, the reference voltage, generated by combining the current and disturbance estimation values from the linear extended state observer, is modulated via PWM and conveyed to the power converter to facilitate torque smoothing control. The efficacy of the proposed control method in enhancing parameter robustness and reducing torque ripple in SRM drive systems has been corroborated through simulations and experimental studies.