Recurrent Neural Network-Based Model Predictive Control for Speed Regulation in Permanent Magnet Synchronous Motors with an Extended Sliding Mode Load Torque Observer

In the pursuit of enhancing control efficiency in Permanent Magnet Synchronous Motor (PMSM) servo systems, this paper integrates a Model Predictive Control Speed Control (MPSC) approach into the speed loop control architecture. The predominant challenge in conventional MPC is its extensive computational demand, which also becomes significantly susceptible to load disturbances. To mitigate this, the paper introduces a discrete recurrent neural network (RNN) specifically aimed at reducing the computational burden associated with MPC. Additionally, an extended sliding mode load torque observer (ESMLTO) is developed for the PMSM drive system, focusing on mitigating load torque effects and bolstering overall system robustness. The viability and performance enhancement brought about by these proposed strategies are substantiated through a series of simulations and experimental validations.