.png){kind=icon}
.png){kind=icon}
Return
Sensorless IPMSM drive with open-loop model-based parameter identification and Hessian-tuned estimator gains
.png){kind=icon}
OA
AI
DOI：10.1016/j.pedc.2025.100131.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This article presents a position-sensorless Interior Permanent Magnet Synchronous Machine (IPMSM) drive scheme that incorporates an open-loop stator current predictor to identify electrical parameters and improve observer performance. Unlike classical observers, the proposed open-loop, full-order model is inherently sensitive to discrepancies between the physical and model parameters. Capitalizing on this sensitivity through current prediction error gradients (PEGs), the parameter estimator scheme identifies the temperature sensitive parameters, i.e., the permanent magnet flux linkage Pm and stator resistance R. An offline experimental method, utilizing the same prediction model, is proposed for mapping the inductance-current relationship, which is subsequently implemented in the processor using a look-up table (LUT). Rotor position and speed are estimated via an Active Flux Observer, with real-time updates to all model parameters. Furthermore, a novel approach for tuning the estimation gain matrix using customized Hessian functions is presented, enabling improved simultaneous identification of Pm and R. Experimental validation is conducted on a 3 kW IPMSM drive, with control and estimation routines implemented on a Zynq System-on-Chip (SoC)based industrial embedded control platform. Results demonstrate that the proposed schemes enhance observer stability, precision, and robustness to parameter variations compared to conventional observers without parameter adaptation.
Keywords:
Hessian
Identification
PMSM
Recursive prediction
Sensorless control
Variable speed drive
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
P
IF:
0
Papers:
19
Citations:
0
Organization
No organization information available