Comparative Analysis of Data Driven Fault Detection using Wavelet and Fourier Transform for Dc Pulsed Power Load in the All-Electric Ship

Next generation warships require energy dense distribution to power advanced weapon and sensor loads. A particularly concerning issue is that these advanced pulsating loads draw large currents in short periods of time and have near fault behavior. This paper compares both wavelet and Fourier transform feature extraction in data driven algorithm. The stationary wavelet transform (SWT) based feature vectors and short-time Fourier transform (STFT) based feature vectors are both fed into a similar statistical data driven fault detection and load monitoring algorithm. The data driven feature vectors corresponding to those transients can be extracted and stored in memory during the training phase of the algorithm. Once the statistical database is populated with those features, each subsequent occurrence of the transient event can be identified. Any shunt fault or disturbance that creates an unexpected event in the load profile will be reliably diagnosed. In final implementation of the feature extraction, the recursive Haar SWT and recursive discrete Fourier transform (rDFT) are used for comparative analysis.