Research of Fault Diagnosis on Explosion-proof Power Capacitors

• Main Authors: CHANG, SSU-TING, 張司廷
• Other Authors: WANG, MENG-HUI
• Format: Others
• Language: zh-TW
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/handle/g6fgb6

碩士 === 國立勤益科技大學 === 電機工程系 === 107 === This study proposes combining the Chaos Theory with an Extension Neural Network for non-explosive power capacitor fault diagnosis. First, the power capacitor faults are classified into five states, the normal state, three-phase unbalance state, aluminum shell expansion, over-temperature, and aluminum shell breakdown. After each state is measured by an oscillograph and the signals are extracted, the chaos dynamic error scatter map is established by using the Chaos Synchronization Detection method, in order to obtain the chaos eye coordinate eigenvalues of different state types, which are identified by the extension neural network algorithm to evaluate the fault type. The experimental analysis results show that the method proposed in this study can effectively identify the fault type of a non-explosive power capacitor, and greatly reduce the data volume of feature extraction, in order to effectively detect the variation of the power capacitor fault signal, and the damage characteristics and operating condition of the three-phase non-explosive power capacitor are known instantly.