Partial Discharge Identification in MV Switchgear Using Scalogram Representations and Convolutional AutoEncoder

This work proposes a methodology to automate the recognition of Partial Discharges (PD) sources in Electrical Distribution Networks using Deep Neural Network (DNN) model called Convolutional Autoencoder (CAE), which is able automatically extract features from data classify different sources. The database used train constructed with real defects commonly found MV switchgear service, and it also includes noise interference signals that are present these installations. PD consist defective mountings, such as loss sealing cap cable terminations, or an earth contact termination insulation. Four were replicated Smart Grid Laboratory on-line measurement techniques obtain signal data. Continuous Wavelet Transform (CWT) was applied post-process into time-frequency image representation. trained predicts high accuracy new data, demonstrating effectiveness partial discharges differentiate them other