Converter Fault Diagnosis Method Based on Principal Component Analysis Combined with Improved Similarity Classifier

Converters play an increasingly important role in the development of microgrids and uninterruptible power supplies. The health of the important components of the inverter will be the focus of future research. This method reduced the data dimension by principal component analysis by sampling the DC converter voltage signal, and selected the data dimension according to the main contribution rate of the preprocessed data. Then, the similarity classifier was employed to identify a variety of different fault types. Finally, the method was used in the dual buck experimental platform. Compared with other methods, the method proposed in this paper has low diagnostic cost, low complexity and high accuracy. Experiment verified the effectiveness and feasibility of the proposed algorithm.