An Online PD Pulses Recognition Method to Extract PD Pulses Buried in External Interferences

Abstract

Partial discharge (PD) pulses recognition, which is directly related to the credibility of PD online measurement results, is the key technology of the high-voltage (HV) equipment PD online monitoring. In this paper, a pulses recognition method during PD online measurement was put forward to extract PD pulses buried in external interferences, the method adopted three key steps to identify PD pulses: (a) Firstly, two pretreatments, including discrete spectrum interferences (DSI) suppression and single pulse boundaries determination, are used to convert several consecutive power cycles on-site monitoring signal to plenty of pulse - time sequences; (b) Then, an optimized adaptive clustering algorithm is adopted to classify all pulses into certain categories (c) Finally, the three-dimensional distribution of each category which indicated the pulses phase distribution with power cycles was calculated, and the PD pulses were identified by the statistics characteristics of their three-dimensional distribution. At present, the PD pulses recognition method has been used in industrial application, and on-site monitoring signal processing results has proved the effectiveness of the method.