Research On State Evaluation And Defect Identification Method Of High Voltage Cable Based On Multi-source Data

Knowing of the operating state of the cable and the discovery of potential defects in the cable timely are the basis for ensuring reliable operation of the power system.Nowadays,many theoretical studies have been carried out on the state evaluation methods of electrical equipment at home and abroad and have achieved good results.However,the weight value given by the normal state evaluation method is fixed and single.It does not consider the influence of the difference of weight and the uncertainty on the evaluation result,and cannot analyze the influence of weight vector deviation on the evaluation results.In view of the above problems,this paper establishes a cable state evaluation model based on weight space sampling,then identifies the cable defect based on the state evaluation results.Because the current defect identification method mainly focuses on the identification of partial discharge mode,but there are other defect types in the cable in actual situation.Therefore,this paper proposes a high-voltage cable defect identification method based on multivariate fuzzy support vector description method from the perspective of data mining.The main research contents of the paper are as follows:(1)Aiming at the problem that the weight value given by the previous state evaluation method is fixed and cannot analyze the influence of the weight vector deviation on the evaluation result,a high-voltage cable state evaluation method based on weight space MCMC sampling is proposed.The set of weight vectors obtained by MCMC sampling is weighted and summed with the degradation degree of the high-voltage cable samples and compared in pairs to obtain the comprehensive deterioration degree result.Based on the probability statistics,the state probability value and the overall priority ranking probability of the object are obtained,and the maintenance scope and the order of the maintenance are determined according to the state probability value and the sorting result.Based on the evaluation results,the cable lines that need to be identified are determined.(2)Aiming at the problem that the discriminant index of cable defect identification method is simple and few studies to identify defects are from the perspective of data mining,the paper proposes a high voltage cable defect identification method based on multivariate fuzzy support vector description.The competitive aggregation algorithm is used to fuzzy cluster and filter the training samples of different states of high voltage cable,and the support vector description(SVDD)is weighted by the membership degree of the filtered samples.Then,the minimum hypersphere model of the training samples of different states of the high voltage cable is established.The state discrimination function is used to identify and judge the defect type of the high voltage cable.(3)According to the data sources of high-voltage cable status comprehensive evaluation and defect identification,including online monitoring data,offline test data,operation and maintenance information,equipment information and environmental and external damage data,a set of cable multi-source status indicators is established.The cables with temperature,partial discharge and ground loop monitoring devices installed in a power supply bureau are used as examples,and the state probability value and sorting result of the object are obtained by the state evaluation method proposed in this paper,and the influence of the randomness of weights on the evaluation results and the ranking results are analyzed.Based on the evaluation results,the effectiveness of the defect identification method is tested by taking the cable with abnormal evaluation results as an example.