PUBLICATIONS

 This thesis presents a convolutional neural network (CNN)-based diagnosis of open switch faults in a dual active bridge (DAB) converter. The operational characteristics of a DAB converter are investigated. Steady-state waveforms are analyzed to identify the operating characteristics of single-switch open faults. Current paths and transformer-side electrical waveforms are investigated under each fault condition. The CNN architecture and the fault diagnosis method for a DAB converter are presented. Faults under various operating and load conditions are learned and classified using the CNN. The validity of CNN-based diagnosis of open-switch faults in a DAB converter is verified through simulation and experiment results.