Online Multiparameter Estimation With Position Error Correction for Unified Synchronous Machine Sensorless Drives

Abstract

For model-based sensorless drives in synchronous machines (SMs), real-time parameter information is essential for accurate position observation. Considering the challenges in fundamental frequency model-based parameter estimation, this article presents an online multiparameter estimation method. The adaptive robust observer in the misaligned synchronous frame small signal model for unified SMs is designed to mitigate the position observation error caused by parameter mismatches. The small signal model, derived from flux linear assumption, effectively accounts for the dq-axis magnetic path asymmetry. With extra small-signal excitation guaranteed by the persistent excitation (PE) condition analysis, the estimated parameters are proven to converge to their actual value under sensorless control within the rated speed. Additionally, a robust feedback function is designed to address the speed fluctuations related to fundamental flux linkage, mitigating their influence on parameter estimation results. For isotropic SMs, the estimated parameters are directly utilized in the model-based observer. For anisotropic SMs, an optimizer is introduced to locate the dq-axis inductance from the coupled inductance matrix in the misaligned synchronous frame. The effectiveness of the proposed method is evaluated through sufficient experiments, showing position observation error within 0.1 rad for IPMSM and SPMSM and 0.15 rad for SynRM.