An Improved Speed Controller for UhsPMSM Based on Simplified Position Estimation and Nonlinear Load Torque Tracking

Due to spatial and speed constraints, sensorless control is a mandatory requirement for ultra-high-speed permanent magnet synchronous motor (UhsPMSM) drive systems. However, conventional position estimation methods struggle to accurately observe AC-form back-EMF at high fundamental frequencies and often involve complex structures. Moreover, the dynamic performance of conventional speed controllers is unsatisfactory due to their limited ability to handle nonlinear load torque variations during speed regulation. To address these problems, this article proposes an improved speed controller to achieve simplified position estimation and nonlinear load torque tracking simultaneously. First, the current dynamic equation is reconstructed in the estimated synchronous reference frame, thereby simplifying the back-EMF to be estimated into a DC component. Second, an explicit relationship between position information and prediction error is derived, enabling direct position estimation through a one-dimensional current prediction equation without the need for arctangent pre-operation. Finally, a cascaded linear extended state observer (CLESO) is designed to estimate position, speed, and load torque, and a speed controller based on estimated load torque feedforward is designed to improve speed control performance. Simulation results demonstrate that the proposed scheme exhibits superior dynamic characteristics during speed regulation, while maintaining comparable position estimation performance to conventional scheme.