Condition Monitoring of PWM Converter Based on Multiple Support Vector Machine Algorithm

Distributed power grid-connected power systems present significant challenges to traditional system protection and control methods. It is the most critical issue for distributed power grid-connected power supply systems to ensure the reliability and security of the system under distributed power access. In this paper, the PWM converter in the power supply system of the radar station is taken as the research object, and the offset law of the midpoint voltage on the DC side of the converter is analyzed. The state changer state monitoring method based on support vector machine is studied. The average value of the midpoint voltage of the converter in the time domain and the voltage energy projection of each frequency band in the frequency domain are used as the fault feature quantity. The simultaneous diagnosis based on the multi-support vector machine algorithm is realized. The insulation balance of the converter and the abnormal frequency of the switching function are verified by simulation experiments.