Feature extraction methods for time frequency energy distribution of PD pulse

Feature extraction methods for time-frequency energy distribution (TFED) using moment, marginal and Renyi information are studied in this paper. They are developed by combination of TFEDpsilas own characteristics with mathematic statistic and information entropy. The Wigner-Ville distribution (WVD) of a simulated PD pulse is chosen as the subject investigated to illustrate those methods, which gives the result of time and frequency moments, time and frequency marginals and Renyi information for the TFED. The results show that those methods can exactly and easily make feature extraction for the TFED of PD pulse, which can be used to PD pattern recognition and in-depth study of PD mechanism in the electrical insulation system.