Fast-switching power semiconductors have higher chance of failure and there is a need to detect and locate these failures to have higher reliability in the shipboard applications. In this work, a deep-learning based fault detection and location method is proposed for open-circuit failure of the switches in 5-level dual-active dc-dc converter. The data driven short-time Fourier transform based feature vectors corresponding to repetitive normal transients can be extracted and stored in the DSP during the training phase of the algorithm for any type of failure. Once the statistical database is populated with those features, each subsequent occurrence of the transient event can be identified. The method is demonstrated using an experimental setup and post-fault switching methods are applied after detecting the fault type.
