Fault diagnosis of power electronic circuits based on rough set-neural network system
Shaofang Wang · Dianli zidonghua shebei · 2004
Based on rough set theory and neural network,a fault diagnosis method for power electronic circuits is presented:rough set-neural network system. The diagnosing process of rough set-neural network system is introduced. Taking tree phase SCR(Silicon Controlled Rectifier) as example,the fault information is processed in advance and the unnecessary fault signs are simplified to form the correct diagnosis rules and to classify the faults. Then the sampled-data of Ud is input to the BP neural network together with the outcome of rough set to locate the fault part. The simulation example proves that the method improves the fault diagnosis speed greatly with high correctness and reliability. This project is supported by the Education Department Fund of Fujian Province(JB01036) and the Ta-lent Fund of Fuzhou University.