Fault Detection in MVDC Power Systems with Wave-let-based Multi-Resolution Analysis

  Copyright: © RWTH Aachen Feature vector extracted from DC current under

Medium Voltage DC zonal distribution architecture is proposed as a new distribution system for the all-electric ship (E-ship) in which the presence of power converters is pervasive. In this context, the detection, location, identification, and isolation of faults within the zonal distribution system will be studied.

In this research, we apply the WT to fault detection in MVDC power systems. WT based analysis techniques have been proposed extensively for fault detection, localization and classification of different power system transients for its capability to extract the abrupt-changing feature of the signals. The MRA is a convenient framework for hierarchical representation of a function at its different scales using discrete wavelet transforms (DWT). In the DWT, the MRA is performed by passing the discrete signal through low-pass and high-pass filters.

The energy contained in the wavelet coefficients of each scale is used as the index for feature extraction. The db10 wavelet and scale 10 are the chosen wavelet function and decomposition level, respectively. The WT method is applied to DC current of the primary bus to detect the faults. The results clearly show an increased energy at the decomposition levels from 6 to 10 of the faulted signal, compared to the normal operating condition. For the next step of this research, a pattern recognition method based on the extracted features, as well as the on-line verification with RTDS and Digital Signal Processor (DSP)will be considered.