An Improved Model-Free Predictive Current Control for SPMSM Drives Based on Adaptive Extended State Observer With Robust Sliding Mode Speed Controller

Conventional ultralocal model-based model-free predictive current control (MFPCC) using an extended state observer (ESO) based on an ultralocal model significantly promotes the robustness against motor parameter mismatch. However, a large bandwidth of ESO leads to an enhanced disturbance rejection capability, but the noise suppression performance will deteriorate. To address this issue, an adaptive extended state observer (AESO) with an adaptive bandwidth tuning scheme based on the estimated error was proposed, in which the observer bandwidth is automatically adjusted to reduce the estimation errors of system states and total uncertainty against the measurement noise. Accordingly, the different requirements of observer bandwidth under various operating conditions can be satisfied, and the performance of disturbance rejection and noise suppression is balanced. Moreover, the stability of the proposed AESO is proved based on the Lyapunov method. Furthermore, to improve the dynamic performance and robustness of the speed loop, a robust sliding mode speed controller (RSMSC) based on a novel reaching law (NRL) and a sliding mode observer (SMO) is proposed to replace the conventional proportional-integral (PI). The comparative experiments on a 1.9 kW SPMSM are conducted to validate the superiority of the proposed control methods compared with traditional methods.