Partial Discharge Gray Image Feature Extraction Based on Bi-directional Two-Dimensional Maximum Margin Criterion

In allusion to the recognition of insulation defect in intermediate joint of XLPE cable,a novel method to extract the feature of partial discharge(PD) grey image based on bi-directional two-dimensional maximum margin criterion(MMC) is proposed.According to the proposed method,firstly based on MMC the obtained PD grey image is projected in two directions,i.e.,the horizontal direction and vertical direction,to extract discriminating vectors for different kinds of grey images,then the nearest neighbor classifier is chosen to classify the PD to recognize different insulation defects in intermediate joint of XLPE cable.The proposed method solves the difficulty of high dimension of grey image feature of PD and less samples to be recognized.The comparative recognition results of PD signals due to four kinds of typical insulation defects of intermediate joint of XLPE cable show that as for the performances in extraction speed of PD feature and recognition rate the proposed method is better the commonly used principal component analysis(PCA) and Fisher discriminant analysis(FDA).