Research On Random Interference With Intelligent Identification Of Partial Discharge Monitoring System

High voltage-bushing is a key part of the transformer,due to the special processing technology of high voltage-bushing,the defects in processing can cause the insulation to degrade and even cause partial discharge.The end result may be that the casing has been damaged by the heat and burst and even the transformer is out of operation,causing huge economic losses.It is necessary to monitor the condition of the high voltage-bushing.Due to the large amount of interference sources in the operating environment of high pressure casing,how to monitor the local discharge signal is a difficult problem to be solved urgently.The on-line monitoring system of local discharge of high voltage casing is based on the high frequency method.The on-line monitoring system includes data acquisition unit,signal processing unit,intelligent control unit,data transmission unit and monitoring host.In this paper,the local discharge signal is collected by monitoring system,and the interference signal in partial discharge signal is studied in order to obtain the accurate discharge of local discharge signal.In order to study the characteristics of jamming signal,the design of the circuit of signal processing unit,intelligent control unit and data transmission unit of the system is emphasized.Besides,independent research and development of high pressure casing of on-line monitoring system for PCB used for real-time display signal real-time waveform,PRPS and fingerprint waveform,moreover also has developed a single pulse capture device,a single pulse spectrum analysis device,and statistical characteristics of device,etc.In this article,we use high-voltage bushing online monitoring system and Schwartz oscilloscope collection devices such as electronic ballast interference,electromagnetic ballast interference,radio interference.By analyzing the waveform,frequency spectrum and waveform,PRPS,and index map of the waveform,PRPS,and map,the characteristics of 7 interfering signals were preliminarily selected.Using MATLAB to train BP neural network,modify the characteristic quantity,and verify the neural network,the identification rate is above 80%,the trained BP neural network is used to monitor the on-line monitoring and control system of high-voltage bushing.Finally,by separating the interference and signal energy,filter the energy of the jamming signal and get the accurate discharge of the local discharge signal.