Research Of Fault Diagnosis System For Capacitive Equipment

The safety operation of the power system is an important problem in the productionlife of the people. In order to ensure the safety operation of the power system, we shouldensure the safety operation of electrical equipment in power system. If we want to preventthe occurrence of equipment failure, we should maintain the electrical equipment regularly.At present, condition monitoring is the trend for preventive of equipment in power system.Condition monitoring monitor parameters of equipment, then detect the fault of equipmentwith the parameters of equipment. So, fault diagnosis of capacitive equipment is the trend ofmaintenance of equipment in power system.The object of this paper is the monitoring parameters of capacitive equipment. Thispaper takes the actual project as its background to detect faults of equipment withmonitoring parameters. So that we can realize the fault detection of the equipment. It takesdifferent methods and the different angles to detect faults of equipment. The issue takesmatlab software and visual studio software as platform to realize fault diagnosis methods.Through the analysis of monitoring parameters, we get the result of fault diagnosis. Weanalyse the data of monitoring parameters to realize the forewarning of abnormal condition.This paper mainly realizes the anomaly analysis and trend analysis of monitoringparameters of capacitive equipment. It realize the fault diagnosis though the following threeaspects:1. This paper applied two layers wavelet tree method for detecting abnormal data onthe dielectric loss factor. This method comprehensive detect mutable burst and persistantburst of abnormal parameter to realize the insulativity of equipment for anomaly detection.2. This paper applied neural network method to monitoring parameters to get therelationship between the multiple parameters and the abnormal situation, then according tothe neural network structure to detect recent anomaly, and to achieve the comprehensivesituation of anomaly detection.3. This paper uses the method of trend analysis to analyze trendency with monitoringparameters, we cut the data segment of monitoring parameters, and then we use the methodof least squares fitting line. According the characteristic we can monitor the abnormaltrendency. Equipment anomalies perhaps reflect in different aspects, we use different kinds ofmethods to detect different types of anomalies, so we can carry out a different point of viewfor fault diagnosis.In this paper, we use above three methods to realize fault diagnosis experiments withthe monitoring parameters of equipment. We can see from the experimental results that thetwo layer wavelet tree method can detect abnormal data in the dielectric loss factor, andthen we can advance warning of anomaly. The neural network method can detect anomalywith comprehensive monitoring parameters, then we can advance warning of anomaly.Trend analysis method can be used to analyze the trend of dielectric loss factor, and then weget abnormal trend, and warn the abnormal trend.