Review of Feature Extraction Methods for Power Equipment Monitoring Data

The breadth and depth of on-line monitoring of power transmission and transformation equipment have been greatly enhanced in the smart grid environment. Time sequence waveform signals are important basis for condition assessment and fault diagnosis of power transmission and transformation equipment, because they occupy a large amount of monitoring data. However, it is difficult to use directly the time sequence waveform signals as machine learning algorithm inputs because of their high dimensionality and large volume. So, feature mining in time sequence waveform signals is the basis and key for subsequent pattern recognition and fault diagnosis. In this paper, the feature extraction is deeply studied for time sequence waveform signal of power equipment monitoring combining with the frequency spectrum analysis and nonlinear dynamics analysis.It intends to provide a reference for further research.