Research On On-line Diagnosis Method Of Cable Accessories Insulation Fault

The increasing demand for electricity day by day puts forward higher requirements for the operation quality of power grid.With the continuous deepening of the transformation of the urban power grid lines,the extensive application of XLPE power cable has brought lots of benefits to safe transmission and improvement of urban appearance.However,accidents are inevitable.Once the cable breaks down,the degree of damage is self-evident.Relevant statistics and research show that the insulation fault of cable accessories has always been the leading cause of accidents in urban grid lines,and partial discharge is one of the early forms of cable insulation fault,which is not only the fundamental cause of cable insulation deterioration,but also the main characteristic quantity of cable insulation.Therefore,the analysis and pattern recognition of cable accessories insulation fault partial discharge signal can timely and accurately grasp the cable insulation status.In view of this,based on the existing research at home and abroad,this paper proposes two online diagnosis methods for cable insulation faults with different targets,and carries out experiments to verify them.The details are as follows:(1)This paper proposes a partial discharge recognition method for cable accessories based on mathematical morphology and fractal theory,aiming at solving the problems of complex recognition process,difficult feature extraction and redundancy selection of cable accessories insulation fault.In this paper,we select the signal in three industrial frequency periods as a sample,and firstly the discharge pulse is extracted by mathematical morphological filtering technology to obtain two statistical characteristics of average discharge and discharge number.At the same time,the Hurst exponent is used to judge the fractality of partial discharge signals.If the condition is satisfied,the box dimension is directly obtained as a fractal feature.Finally,the combination features are constructed and imported into the extension neural network to complete pattern recognition.(2)This paper proposes a partial discharge recognition method for cable accessories based on convolutional neural network under small data volume,aiming at solving the problem that convolutional neural network requires large data set in the application of cable accessories insulation fault type recognition,and deeply mining the data characteristics.PD signals have chaotic characteristics.In this paper,equivalent transformation of partial discharge signals is carried out through phase space reconstruction to make its chaotic characteristics more obvious.Then the combined features,including geometric feature,chaos feature,entropy,fractal feature and time domain feature,are extracted to increase the volume of characteristic data.Finally,the convolutional neural network is introduced to complete pattern recognition.Simulation and test results show that the partial discharge recognition method for cable accessories based on mathematical morphology and fractal theory proposed in this paper simplifies the flow fault diagnosis,and only three features can be extracted to complete the fault diagnosis.The partial discharge recognition method for cable accessories based on convolutional neural network under small data volume can fully extract data features,and can also complete the fault diagnosis when the amount of data is small.Both methods effectively make up for the shortcomings of the existing similar methods in feature extraction,method flow and other specific aspects,which has high practical application value.