Partial discharge classification in 11 kV XLPE cable joints using artificial intelligence based techniques
A condition monitoring (CM) system, which integrates wavelet analysis and artificial intelligent techniques to analyze the partial discharge (PD) patterns and classify the types of PD in 11 kV 3 core XLPE cable insulation is developed in this work. Several types of defects are artificially introduced in the cable insulation to create different types of PD (void, surface and corona). Wavelet transform is employed to de-noise the PD signals. The performance of several well tried wavelets is investigated. Different feature extraction methods are applied to the de-noised PD signals to form pattern vectors, which are used as the inputs to a proposed neural network based classifier to identify the types of PD patterns. Lastly, the performance of the proposed CM system is verified with another sets of data obtained from different PD sources.
Date of Project: September 2013 - August 2016