A Comprehensive Method for Online Switch Fault Diagnosis and Capacitor Condition Monitoring of Three-Level T-Type Inverters

Capacitors and semiconductor power devices are the most fragile components in power electronic systems. Hence, designing a comprehensive method for online switch fault diagnosis (FD) and capacitor condition monitoring (CM) is conducive to guaranteeing the reliability of the three-level T-type (3L-T) inverter in an all-around way. First, based on modeling the 3L-T inverter on the dc link and ac side, the inherent harmonic components of the dc link under normal and open-circuit fault conditions are analyzed. Furthermore, the capacitor current reconstruction method is designed. Subsequently, a comprehensive FD and CM framework is proposed, which consists of signal injection, extra hardware, 3L-T model, FD, and CM submodules. In FD, the detection variable, which is the function of output current and its changing rate, is designed, and the fault location is carried out based on the phase current residual and the sum of the neutral-point voltage residual. The capacitance estimation method is introduced based on the input current reconstruction technology and recursive least squares algorithm using the inherent or injected ac component on the dc link. Finally, experimental results demonstrate that the proposed comprehensive method merits high estimation accuracy, fast dynamic response, and better FD performance under various conditions.

Automated summary

This paper proposes a comprehensive method based on modeling to achieve online switch fault diagnosis and capacitor condition monitoring in order to ensure the reliability of the three-level T-type (3L-T) inverter. The inherent harmonic components of the dc link under normal and open-circuit fault conditions are analyzed, and a capacitor current reconstruction method is designed. A comprehensive framework for fault diagnosis and condition monitoring is proposed, which includes signal injection, extra hardware, 3L-T model, fault diagnosis, and condition monitoring submodules. Experimental results demonstrate that the proposed comprehensive method achieves high estimation accuracy, fast dynamic response, and better fault diagnosis performance under various conditions.