PUBLICATIONS

This paper presents an open-switch fault detection for a Vienna rectifier using a convolutional neural network (CNN). Open-switch faults are diagnosed by analyzing the output characteristics of the system. The output characteristics consist of the output voltage and the input current. To accurately identify the location of the open-switch, both the voltage and current are analyzed. The current conduction paths under various open-switch fault conditions are analyzed prior to applying the CNN, in order to understand the behavior of voltage and current distortions. Open-switch faults distort the waveforms of the output voltage and input current, and these distortions are extracted using CNN composed of multiple convolutional and fully connected layers. The CNN is trained to classify the location of the open-switch based on the voltage and current data. The feasibility and effectiveness of the proposed method are verified through both simulation and experimental results.