Research On Insulation Condition On-line Monitoring And Fault Diagnosis Of Crosslinked Polyethylene Power Cable

Cross Linked Polyethylene(XLPE)cable is a key carrier for power transmission and plays a crucial role in the transmission network and the whole power system.The insulation performance of XLPE cable greatly affects the stability and safety of power grid operation and determines the safe and reliable operation of the whole system.As a precursor phenomenon and main cause of cable insulation aging,partial discharge can effectively characterize the insulation health of XLPE cables,so it is important to carry out research on online monitoring and fault diagnosis technology of XLPE cable insulation condition based on partial discharge to grasp the insulation condition of cables and ensure the stable operation of power systems.This paper focuses on the online monitoring and fault diagnosis technology of XLPE cable insulation condition.First of all,the insulation condition monitoring should be carried out on the basis of in-depth research on the fault mechanism and partial discharge mechanism of the cable.To this end,firstly,a detailed study of XLPE cable structure is carried out,and typical insulation faults and their causes are statistically analyzed;then,the insulation fault mechanism is summarized in conjunction with five types of aging effects that cause cable insulation faults;finally,the mechanism of partial discharge is analyzed in depth,and the discharge characteristics of different partial discharge types and the hazards of partial discharge on cable insulation media are summarized.Secondly,the implementation of accurate insulation condition monitoring is based on the comprehensive and effective acquisition of local discharge signals.XLPE cable operating environment is complex and the local discharge signals in the field usually contain a large number of noise signals,for this reason,this paper proposes a local discharge signal denoising method based on adaptive variational modal decomposition.Then,based on the idea of wavelet threshold denoising,the wavelet threshold function is derived and improved to improve the denoising performance of wavelet decomposition method.Finally,the results show that the proposed method can not only achieve the effective separation of signal and noise,but also retain the signal characteristics intact.Finally,the accurate identification of partial discharge patterns is a prerequisite for fault diagnosis of cable insulation conditions.The CNN-Transformer combines the advantages of local feature extraction ability of convolutional neural network and Transformer’s good at mining temporal context features,and uses local discharge signal phase pattern as the input.The CNN-Transformer combines the local feature extraction ability of convolutional neural network and the advantages of Transformer in mining temporal context features.The experimental results show that CNN-Transformer has better recognition performance compared with the traditional method.