| Summary: | 碩士 === 國立臺灣科技大學 === 電機工程系 === 98 === In order to improve the quality of distribution service, security and appearance, the traditional overhead distribution system have gradually evolved into underground distribution system. With all the factors would lead to the deterioration of electric cable insulation materials. Furthermore, the common accidents of underground cables usually take place in the artificial connecting part. Therefore, this paper interprets underground cable joint insulation status diagnosis.
The thesis draws on the test data of underground cable joint to make real-time insulation diagnosis. Making an artificial defect at the joint connection point and applying high voltage to it until the material breakdown. The experiment voltage and signal of partial discharge are measured at a fixed time interval during experiment. The PD signal will be converted into several time-series characteristic parameters by filtering and simplifying, and then the sample data will be acquired with fixed width by it. Respectively using Normal distribution, Log-Normal distribution, Gamma distribution and Weibull distribution to execute goodness-of-fit test with sample data to verify which kind of probability distribution is suitable.
Through the distribution trend of the measurement data, the highest rate of compliance test is gravity center of discharge. As a result, the characteristic parameter is the insulation representative. From the result of observing each trend of characteristic evolution, the characteristic parameter will appear outliers when the measurement data is with gap. The positions of outliers recorded with third standard deviation threshold are taken as the basis of initial period, middle period and final period of real-time insulation status diagnosis.
|
|---|
