Inverter Open Circuit Failure Detection Method via Residual Inner Product Feature

In view of the problems of weak decoupling ability of uncertain signals and difficulty in extracting fault features in current failure detection methods, an open circuit failure detection method of inverter based on residual inner product feature is proposed in this paper. Firstly, the three-phase current noise filtering method based on robust differentiator was proposed for subsequent residual generation and failure detection. Secondly, the mathematical model of inverter based on generalized state average method was constructed under healthy working conditions. The current residuals were obtained by subtracting the current model values from measured fault values. Thirdly, by defining the inner product operation, the residual inner product features under each time window were calculated successively. The alarm threshold with weights was calculated. Failure detection is achieved by comparing the two. The failure detection time and specific failure phase are obtained. The accurate failure power transistor location is achieved by combining the current residual polarity.