A New Method Of Fault Location For Transmission Line Based On Artifical Neuro Network

Fault location for high voltage transmission lines has great social and economical benefit, now electric manufacturers and scientific researching organizations are doing work- about it.Fault location for high voltage transmission lines is influenced by many factors such as fault transient impedence, load currents and source impedence. It is hardly depicted in accurate mathematics model, and it makes traditional methods of fault location defective and insufficient. Artifical neuron network has a strong ability of model categorizing, nonlinear mapping and association. Now, it has become researched hotpot in the fields of relay protection and fault location.Because of the shortage in fault location using single terminal data and fundamental function of ANN, a new faultlocation method has been presented for transmission lines based on ANN. It is supported by the fact that the fault distance and fault data has homologous relationship.First, on the basis of the theory of MIL (Multiconductor Transmission Line), the approach of analysis in s-domain and the technique of Hobson' s numerical inverse Lap lace transform, the accurate computation of transmission lines fault electromagnetic transients is presented by which fault-transient data is produced for training and test ANN.Secondly, a series of experiments has shown that singular ANN for fault location is neither convergent after long training nor able to practical application. Furthermore, the ANN is morbid. On the basis of singular Anna ANN of delaminating & distributing is given which includes two layers. In the first layer, five kinds of transmission line faults are compartmentalized. The second layer has five sub-ANN, which can performs the task of accurate fault location.Finally, in order to examine the synthesized performance of fault location, a great lot of fault swatches are input into the ANN .The result shows that this new method can conquer or eliminate the influence of fault transient impedance and source impedance, so it is efficient and viable with higher precision of fault location.