A Fault Diagnosis Method Used for the Three-Phase Full-Bridge Rectifier Circuit

Abstract

When talking about three-phase full-bridge rectifier circuit fault diagnosis, the ordinary neural network diagnosis shows its limitation on local minimum point in training process, and low diagnosis accuracy. Moreover, this kind of fault diagnosis is based on fuzzy logic and it relies too much on experience of expert through trial and error method, without self-learning capability. This paper provided another solution on theoretical analysis and research of fault diagnosis, which was named as From Vector to Subspace algorithm. It is a time-domain analysis method to detect fault location: Firstly, classifying thyristor faults, simulating the target circuit with MATLAB Simulink and obtaining load voltage signals, Secondly, extracting fault features from the signals and defining the fault space, then taking use of algorithm with new fault feature and their spaces, to get final results. After those, it was approved that: this kind of time-domain fault diagnosis algorithm has a number of advantages such as low complexity, small amount for computation, intuitive, fast and efficient, and higher rate of fault diagnosis. With this algorithm, the fault classification problem of three-phase full-controlled bridge rectifier circuit could be easily detected and solved, which would help engineers to understand states for electronic devices.