Fault Diagnosis of Three Level Inverter Based on Multi Neural Network
Chen Danjiang1,2,Ye Yinzhong3
1. Shanghai Maritime University Shanghai 200135 China 2.Zhejiang Wanli University Ningbo 315100 China 3. Shanghai Institute of Technology Shanghai 200235 China
Abstract In order to diagnose different fault patterns of single device open circuit and several devices open circuit at the same time,a fault diagnosis method for a kind of three level inverters was discussed in this paper. Due to the reason of some fault patterns can not be isolated easily,the upper,middle and down bridge voltages were used as test signals. Then the fault features including harmonic amplitude and phase were extracted from these signals use the method of spectrum analysis. At last,a novel multi neural network was proposed to diagnose all these open circuit faults. The simulation and experiment result show that the method can isolate fault patterns exactly. This method has the advantages of high classification performance and high reliability.
Chen Danjiang,Ye Yinzhong. Fault Diagnosis of Three Level Inverter Based on Multi Neural Network[J]. Transactions of China Electrotechnical Society, 2013, 28(7): 120-126.