Partial Discharge Detection And Discharge Type Identification Of Insulation Defects

With the increase of the power system capacity,the applied electric field to the power equipment is increased correspondingly.Insulation defects will inevitably occur during the manufacture and operation of the power equipment.Partial discharge will occur firstly in these defective parts.Further enlargement of the partial discharge will lead to the failure of power equipment,thus giving rise to a serious threat to the safe and stable operation of the power system.Therefore,the detection of partial discharge and the correct identification of the type of partial discharge are of great significance to the normal operation of power system.After studying the background and significance of the topic,the mechanism and detection method of partial discharge are researched in this paper.Through the MATLAB/Simulink simulation analysis of single air gap insulation defects,we have a deeper understanding of the partial discharge process.The paper are carried out from the following three aspects: the partial discharge test,the acquisition and processing of partial discharge data,and the pattern recognition of the type of discharge.In the part of partial discharge test,the sample models based on four typical types of insulation defects commonly found in power equipment are designed and produced.The pulse current method is used to detect the partial discharge.The simulation to determine the parameters of the detection impedance is performed with ATPDraw.According to the basic circuit of pulse current method,a partial discharge test platform is built in the laboratory to test the insulation defect samples.In the acquisition and processing of discharge data section,a set of data acquisition and processing system based on LabVIEW is developed.The system includes four modules: system configuration,data acquisition and storage,data viewing and playback,and data processing.The four modules are respectively designed and developed.After the data is processed,the discharge spectrum of partial discharge corresponding to different types of defects are obtained.In the part of pattern recognition of discharge type,the skewness,kurtosis and cross-correlation of the discharge signal are selected as the discharge characteristic parameters and extracted as the input of the pattern recognition network.The BP neural network is used as a pattern recognition network to identify the type of discharge,and a simple BP neural network is optimized through additional momentum method and variable learning rate method.The BP neural networks before and after optimization are trained and tested separately.The results show that the data acquisition and processing system developed in this paper can achieve the acquisition and processing of partial discharge data in insulation defects successfully.Using the extracted characteristic parameters as input,the optimized BP neural network can identify the type of discharge effectively.Through the optimization of additional momentum method and variable learning rate method,the recognition rate of BP neural network is improved.The design of this subject has reached the expected goal.