A Interval Diagnosis Approach of Soft-Fault Diagnosis for Large-Scale DC Analog Circuit

In order to resolve the difficult of soft-fault diagnosis in large-scale DC analog circuits,a soft-fault diagnosis approach,which is based on network decomposition,fault state description by interval and fuzzy neural networks(FNN),is proposed.First,when potential fault component changes its parameter in global scope,the voltage interval of every measurable node which belongs to the subnetwork is emulated out.Then,these voltage intervals are divided into several subintervals based on the tolerance of component parameter,thus the complete description of fault state is achieved.With the difficult in resolving interval data by neural networks,this paper preprocesses input signal with fuzzy algorithm,and then locates fault component by BP neural networks.Finally,the efficiency of the method is shown by a practical example.