Robust Active‐Flux Hybrid Observer With an Estimated dq‐Axes BEMF Observer for IPMSM Sensorless Control
Chao Yao, Sheng Huang, Dacheng Ni, Liangxing JiangABSTRACT
Parameter mismatch often induces a non‐negligible position error in active‐flux‐based hybrid flux observers for sensorless IPMSM drives. This paper first analyses the position‐error mechanisms caused by stator‐resistance, q‐axis‐inductance, PM‐flux‐linkage and d‐axis‐inductance mismatches, revealing their distinct sensitivity characteristics with respect to speed and load. To compensate the mismatch‐induced position bias, a reduced‐order back‐electromotive‐force (BEMF) observer is designed in the estimated synchronous reference frame defined by the original uncompensated position and speed. The position‐error information embedded in the estimated BEMF components is extracted and used to correct the original angle estimate online. The internal stability of the reduced‐order observer, gain‐selection guideline, linear angle approximation, and parameter‐mismatch sensitivity are further analysed. Experimental results under load steps, speed decrease, speed reversal, wide‐range acceleration/deceleration, ultra‐low‐speed operation and single‐parameter mismatch tests demonstrate that the proposed method reduces position error and load‐induced speed sag, improves low‐speed operating stability and achieves competitive robustness compared with an ESO‐based compensation method.