Noncommunication protection of transmission line based on genetic evolved neural network

This paper first introduces a novel fault transient measurement arrangement, and the simulation studies relating to faults on typical 400 kV vertical construction transmission lines to examine and analyse the fault generated current signals. The application of the ANN based fault measurement technique is then described. An Evolving Connectionist Network (ECN) with a GA is investigated to solve the problem of internal and external fault identification. The technique is based on utilising fault generated high frequency current signals to essentially recognise the various patterns generated within the frequency spectra of the signals, for the purposes of accurately distinguishing between internal and external faults under various system and fault conditions. In particular the effect on the higher frequency current signals (and therefore on the performance of the ANN-based protection technique) due, for example, to fault type, fault position and fault inception angle for nonlinear arcing faults is highlighted. The paper clearly demonstrates that the new ECN method can successfully optimise a connectionist network for fault identification in EHV transmission lines thereby significantly improving performance over backpropagation learning algorithm based NN