Real-time Insulation Status Assessment of XLPE Based on Decision Tree Scheme

• Main Authors: CHING-MING CHEN, 陳慶明
• Other Authors: Ruay-Nan Wu
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/02198731968571542915

碩士 === 國立臺灣科技大學 === 電機工程系 === 98 === 　　The cross-linked polyethylene(XLPE) is the general insulation materials for underground cable. However, it might break down when the cable start to insulation deterioration and then will cause the economic loss and the damage. To perfectly prevent the damage of underground distributed system, this research set up a set of on line monitoring system of partial discharge(PD) to prevent these damages happened. 　　This thesis is about the slices test of XLPE and made the high voltage and accelerated deterioration test in high voltage laboratory in NTUST. The next step is that to filter and simplified the signal and going to the data switch and further to use feature extraction methods to figure out the feature data of insulation status. This research uses three kinds of feature data that the characteristics are the high Similarity of district average discharge of each test data and the turning properties of slop curve and the stair rises properties of interval extremum deviation. The final step is to use the decision tree scheme to organize the district average discharge, slope curve and interval extremum deviation three feature data and then find the if-then rules further more to set up the on line insulation state diagnosis rules. 　　The decision tree consists of six decision nodes and ten leaf nodes. The accuracy of real-time insulation status diagnosis about testing data are 95.6% during early stage of degradation、89.8% during middle stage of degradation and 81.3% during final stage of degradation.