The pattern recognition of neural network is applied to the fault location for transmission line. The whole structure is composed of fault detecting subnet, fault phase selecting subnet and fault locating subnet. When fault detecting subnet detects fault, fault phase selecting subnet goes into operation to ascertain the type of fault. At last, fault locating subnet exports the distance of fault. The pretreatment of input data makes use of the combination of wavelet transform and energy entropy. Various fault simulations to the selected power system model using Matlab. Simulation results shows that fault location based on BP neural network is free from the disturbance of fault type, fault resistor, fault location and potential angle.
|