2026/06/01 by Wenbing Hu, Jiayan Zhang, J. Zhang +7
Engineering · #Fault Detection and Control Systems #Machine Fault Diagnosis Techniques #Magnetic Field Sensors Techniques
paper · doi:10.1109/tpel.2026.3697744
openalex publication_date 2026/06/01 · openalex created_date 2026/06/02 · openalex updated_date 2026/07/29
This paper proposes a hierarchical fault-isolated observer (HFIO) framework for coordinated incipient current sensor (CS) fault diagnosis and sensorless control in permanent magnet synchronous motor (PMSM) drives. The core challenge lies in the conflict between robust sensorless state estimation and sensitive diagnosis of weak incipient CS faults, since conventional observer-based schemes typically rely on raw phase-current residuals that remain strongly coupled with the control loop and are vulnerable to open-circuit (OC) fault disturbances. To address this issue, the proposed HFIO framework introduces a dual layer heterogeneous architecture. In the control layer, an adaptive variable-exponent high-order terminal sliding-mode observer (AVE-HOTSMO) is designed to balance transient convergence and steady-state chattering suppression. Simultaneously, in the diagnostic layer, a velocity-scaled open-loop state observer (VS OLSO)generates a decoupled residual in which incipient CS fault signatures remain stable and distinguishable from OC-induced disturbances. The steady-state fault evolution in this residual is analytically derived under two-phase current sensing, revealing a rotational fault-signature mapping that forms a complete model driven fingerprint library. Based on these signatures, a hierarchi cal diagnostic strategy is established, combining adaptive-baseline dual-window statistical difference (ADW-SD) for robust anomaly detection and state-gated sequential probability ratio test (SG SPRT) for high-confidence fault localization. Experimental results verify that the proposed framework achieves accurate sensorless control and reliable incipient CS fault diagnosis with strong robustness in complex dynamic environments.