Monitoring Partial Discharges Of GIS And Identifying The Type Of Faults

Higher and higher quality electrical Power supply is needed with the development of the society. So gas insulated switcher (GIS) has more and more applications in the power system due to its characteristics advantages, such as small volume, high quality, and so on. Essentially, GIS is designed not to need maintenance. However, in case of an insulation breakdown, it takes a longer time for its restoration and breakdown causes more serious damage to system operation.As is known to all, partial discharge (PD) is a symptom of an insulation breakdown occurring in GIS. So it is very important to develop a new diagnostic technique to detect a characteristic symptom when PD occurring. At present, signals of light, supersonic wave, electromagnetic wave emitted from GIS can be utilized to determine whether PD occurs, however, the results are not satisfying due to the processing methods and the strong noise at the sites. To solve this problem, this paper proposes the method, which is most popular used the world over, called pulse-current, also be recommended by IEC270, to receive the unmoral signals from the equipment. The radio detectors can acquire the voltage wave. By through the wave filter, we can obtain the purified unmoral signals caused by the electric equipment.This paper focuses on the identification of partial discharge by GIS. The method includes frequency determination based on FFT, equivalent time-frequency method, fuzzy C-means clustering analysis, the peak-phase mapping method, support vector machine, wavelet analysis and fuzzy identification and so on. The algorithms combine together to detect partial discharge of GIS (Gas isolating Switch) detection methods and recognition technologyWhen the type of fault is more than one or the interference is excessive, by using the characteristic factor calculated by traditional method of peak-phase spectrum directly will lead to great inaccuracy for identification, will have great difficulties. Equivalent time-frequency method has the ability to classify abnormal signals effectively, so it is easy to identify the classification of different kinds of signal. It is possible to make the same type close together, and then fuzzy C-means theory can detach the same type of signals, thus the probability that the next identification process can be processed exists. In the light of the traditional method using phase characteristics for identification, the feature parameter that includes more than ten types and dozens of numbers, thirteen of which are chosen to the characteristic vector. In the end of the step, the identification process is completed by support vector machine (SVM) belonging to artificial neural network (ANN).In order to acquire a better determination con conclusion of discharge type, and with the purpose that detection system can be easily used for other electrical equipments and testing equipments, DC equipment for instance, the system increases identification mode of point-by-point. Using the wavelet analysis of signal energy reflect single signal's frequency, time, and the peak of the energy. Then the energy distribution characteristics of signal could be used as recognition characteristics. Utilizing wavelet analysis of energy information as features, then use fuzzy identification method to determined a single signal types. This paper taking bus discharge, internal discharge, the most common type as examples, illustrates two kinds of thesis combine together for detection and recognition process. Simulation experiments show that the system can effectively distinguish discharge type.