Complex-Vector Isotropic Nonsingular Fast Terminal Sliding Mode Observer for High-Precision Speed Estimation of Sensorless Surface-Mounted Permanent Magnet Synchronous Motor Drives

Sensorless position and speed estimation methods for permanent magnet synchronous motors (PMSMs) eliminate the need for mechanical sensors, thereby reducing costs, enhancing reliability, minimizing system volume, and enabling maintenance-free operation. Furthermore, these approaches achieve high-precision control and exhibit strong adaptability to harsh environments. Particularly in medium- and high-speed operations, sensorless schemes based on the nonsingular fast terminal sliding mode observer (NFTSMO) offer strong robustness and rapid convergence. However, inherent speed chattering increases speed fluctuation and introduces adverse harmonic distortions in the estimation system. To address this, this paper proposes an improved observer strategy based on complex-vector reconstruction. This method formulates a reduced-order system model in the complex domain and constructs an isotropic sliding-mode surface, effectively suppressing the odd harmonics induced by the algorithmic structure. Additionally, a nonlinear tracking differentiator (TD) is introduced to filter out high-frequency noise from the estimated position and speed signals. Simulation and experimental results demonstrate that, compared to the conventional NFTSMO, the proposed method mitigates high-frequency chattering and improves speed estimation accuracy in medium- and high-speed sensorless operations, while maintaining comparable position tracking accuracy and convergence speed.