Research On Partial Discharge Pattern Recognition Of GIS Based On UHF

Modern power system is facing a large power grid,large units,high voltage,large capacity direction,in order to protect the stability and reliability of power systems,higher requirements are put forward for electrical equipment.While the traditional power transmission equipment and power conversion equipment can not meet the requirements in terms of reliability,security and flexibility of modern power grid operation.Since the beginning of the eighties of last century,gas-insulated switchgear(GIS)equipment has been widely used in power transmission and converter stations because of its small size,small environmental impact,long maintenance period and high operational reliability.Today,GIS has become the mainstream of modern power system power transmission equipment.At the same time,due to GIS equipment in the manufacture,transportation,installation and operation of various uncontrollable reasons,partial discharge caused by a variety of different types of defects within GIS will occur.The long-term practice has proved that the partial discharge detection technology can detect the local discharge phenomenon,the defect information and the severity of the GIS equipment in advance,and also can realize the accurate positioning of the defect,and make the warning of the fault,so as to realize the planned maintenance and reduce the equipment damage and the occurrence of accidents,it is of great significance for the stability and safe operation of the power system.In this paper,the definition and generation of partial discharge are introduced briefly.The main partial discharge detection methods of GIS equipment are classified.Then,the design of the GIS experimental device,the layout of the equipment and the setting of the wiring are introduced in detail of XD5936,which is produced by Hangzhou West Lake Electronic Research Institute,is used as the test platform,and several different detection methods are presented in this paper.GIS equipment detection methods were briefly introduced.Then,the model design principle and local discharge mechanism of four kinds of typical defects of XD5936 experimental device are studied.Lastly,four typical flaw types of partial discharge waveforms are collected by UHF detection method in GIS devices for subsequent studies.The collected signals are first pre-processed,and then more than 30 characteristic information parameters of GIS UHF detection signals are extracted based on statistical information,time domain information and frequency domain information.In this paper,multi-envelope feature extraction method of the the time domain signal is proposed creatively,the method is easy and specific,the redundant information in the detection signal is removed,at the same time eliminating the interference of random signals to a certain extent.Then,the feature extraction of the frequency domain information is extracted by envelope.(LLE)algorithm is used to reduce the dimension of 36 feature quantities,and finally 10,16 and 20 three different combinations of characteristic parameters are obtained by setting the parameters respectively.Lastly,RVM,as a classifier,is selected for it's more generalization ability,the more sparse solution and more robust,and constructs a classifier called M-RVM,which can be used for multi-class prediction and classification by "one-to-one" method combination of RVM,and then select the above three different feature reduction dimension combination results as the input vector of M-RVM classifier for classification and prediction comparison,the results show that the dimensionality reduction of 16 get best result.At the same time,by comparing with the classification method of the neural network,the results show the identification system which combination LLE reduction algorithm and the M-RVM classifier has reliable identification for the partial discharge detection caused by the defect type of the GIS device.And it is proved that this method has certain guiding significance and feasibility in detecting the insulation condition of electrical equipment,understanding the internal insulation defects and preventing latent and sudden accidents.