断层(地质)
国家(计算机科学)
控制理论(社会学)
计算机科学
电子工程
工程类
算法
控制(管理)
地质学
人工智能
地震学
作者
Benchao Zhu,Lei Wang,Zichen Gao,Minsheng Zhou,Lijun Diao
标识
DOI:10.1109/tim.2025.3599275
摘要
Fault diagnosis is crucial to ensuring the reliability of permanent magnet synchronous motor (PMSM) drive systems. For PMSM drive systems utilizing three-phase two-level inverters, power switch open-circuit (OC) faults are one of the most common failure modes. However, most existing switch OC diagnostic methods are limited to diagnose single switch OCs, are unable to identify simultaneous double switch OC fault, and necessitate extensive experimental calibration of their diagnostic thresholds. This article proposes a unified method capable of diagnosing single or double switch OC faults. This approach divides the post-fault response into distinct states based on changes in current paths, where each state is identified by assessing whether the residue between the inverter’s expected voltage and the actual output voltage exceeds a specific threshold. Furthermore, the localization of single or double switch OC faults is achieved by tracking the transitions among these fault response states. To enhance robustness, adaptive thresholds are formulated using the median of the observer error and equivalent input derived from a current sliding mode observer. One threshold serves to detect fault occurrence, while the other facilitates OC localization. The proposed method requires no supplementary hardware and demonstrates robustness against variations in control parameters, diverse operating conditions, and sensor noises. The effectiveness of the proposed method is validated through extensive experiments.
科研通智能强力驱动
Strongly Powered by AbleSci AI