Feature Extraction using Symbolic Dynamic Filtering for Fault Analysis in Distribution Systems

Feature extraction for fault detection and classification using Symbolic Dynamic Filtering (SDF) is explored in this paper. It provides an edge over existing methodologies by compressing voluminous waveform data into probability histograms which describe signature features of the faults. A SDF is constructed based on the knowledge of symbolic encoding and finite state automata to generate signature histogram patterns from different fault categories. These histogram patterns are used for training various pattern classifiers to build the fault classification model. The simulation results on a test distribution network showcase the model accuracy for fault prediction with k-nearest neighbour (kNN) and support vector machines.