Study On Partial Discharge Fault Identification In GIS Based On EEMD Energy Moment And ISSA-SVM

With the wide use of gas Insulated Switchgear(GIS)in power transmission and distribution systems,it has been recognized for its good insulation performance,high power supply reliability and small footprint.However,due to the growth of GIS devices in the power system,the faults caused by the daily assembly,transportation,operation and maintenance of GIS devices are also increasing.Partial Discharge(PD)caused by different types of insulation defects has different hazards to the insulation of GIS devices and different hazards to breakdown.It is of great significance to identify the type of partial discharge for the treatment and maintenance of GIS equipment.However,due to the diversity and complexity of partial discharges,it is difficult to identify the type of partial discharges by a single mode analysis,and the results are not good in practice.Therefore,in order to identify the type of partial discharge efficiently and accurately,and to ensure the safe and stable operation of the device,a method for identifying the type of partial discharge in GIS based on Ensemble Empirical Mode Decomposition combined with the improved Sparrow Search Algorithm-Support Vector Machines algorithm is presented.First,the partial discharge model is designed,the partial discharge signal is collected,and the EMD algorithm is used to test the original waveform decomposition.Then,a set mode decomposition EEMD method is presented based on the empirical mode decomposition method.The EEMD algorithm is used to test the initial waveform decomposition to prove the superiority of EEMD in signal decomposition.The data are processed by the energy moment.Secondly,SVM algorithm is used for pattern recognition of discharge signals.However,the SVM algorithm needs to optimize the parameters of the kernel function in order to improve the accuracy,so based on the Sparrow Search Algorithm,an algorithm ISSA based on the cosine function for updating the proportion of individual explorers is proposed,and its performance is tested using the particle swarm optimization algorithm PSO and the SA algorithm respectively.Proves the superiority of ISSA in search performance.Then the PSO-SVM,SSA-SVM,ISSA-SVM mathematical models are established,and the performance tests are carried out using the wine dataset to verify that ISSA-SVM is better than the other two commonly used methods in data processing.Finally,an experimental platform for GIS partial discharge which can produce four types of partial discharge is set up to get four kinds of partial discharge signals.Then,four kinds of partial discharge signals are separately decomposed by EEMD combined with energy moment algorithm,and their feature vectors are extracted.The optimized SVM algorithm by ISSA algorithm is used to identify the types of GIS partial discharge.The results show that the proposed method can effectively identify different types of partial discharge in GIS,and improves the recognition accuracy by 16.7% and 8.5% compared with PSO-SVM and SSA-SVM algorithms,and requires fewer iterations.The comprehensive diagnostic accuracy reaches 96.25%,which verifies the effectiveness and superiority of the proposed method.This paper has 45 figures,7 tables and 60 references.