Research On Online Detection Of Partial Discharge And Defect Identification Of Power Cables

With the continuous development of urban and rural modernization,the power industry plays an important role in people’s daily life.Underground power cables play an important role in power transmission and distribution,and are increasingly favored by the power industry.There will be some faults in power cables in daily operation.One of the main reasons for these faults is that the insulation layer of power cables is prone to partial discharge.The strength of partial discharge directly affects the insulating ability of the power cable insulation layer,so detecting partial discharge of power cable is one of the effective methods to prevent accidents.Due to the limitation of inability to monitor the discharge state of the cable in real time by the traditional method of regular or power outage maintenance,the electromagnetic coupling method and the pattern recognition of defects in the on-line inspection of power cables are deeply studied.Firstly,the research background and significance of the subject are briefly introduced,and the domestic and foreign development status of power cable partial discharge on-line detection technology and the development process of defect pattern recognition technology are studied.Taking the air gap in the insulation of the power cable as an example,the occurrence of partial discharge and its periodic causes are analyzed.Compare the advantages and disadvantages of various power cable online detection methods,and use the electromagnetic coupling method for partial discharge online detection.Secondly,the working principle of the Rogowski coil in the high-frequency current sensor and the precautions during production are briefly described,the coupling signal ability of the high-frequency current sensor at different frequencies is tested,and the combination of software and hardware is used for anti-interference.And four kinds of defect discharge models are made,and an on-line partial discharge detection system based on electromagnetic coupling method is built.In terms of signal acquisition,the combination of data acquisition card with FPGA as the core and computer is selected;in terms of software,Matlab software programming is selected to perform data extraction and noise reduction processing,generate corresponding discharge spectrum,and analyze each type of discharge.Finally,the method of support vector machine is used for partial discharge defect pattern recognition,the classification principle of support vector machine is studied,and radial basis（RBF）kernel function and hyperbolic tangent（Sigmoid）kernel function are used to construct the classification function of high-dimensional space.In order to correctly identify different discharge types,statistical feature parameters are selected to perform feature extraction on partial discharge data,and this feature parameter is used as the input feature vector of the classifier.The principles and advantages and disadvantages of the three classification decision-making methods are analyzed,and a one-to-many classification algorithm is selected to construct a classifier.Through continuous training and optimization of the classifier,the pattern recognition of partial discharge is carried out.Through the comparison of the recognition results,it can be seen that the recognition accuracy of the RBF kernel function is above 90%,and the recognition effect is more accurate,which can be used in the classification and recognition of cable partial discharge defects.