Transmission Line Fault Recognition Based On Elman Neural Network

The power station and the electricity consumers is connected by transmission lines, it’sthe main artery of power system. Transmission lines are generally set up in the remotemountain area and harsh environment for the safety and convenience of residents’ life, sotransmission line faults are happened frequently. Once a fault of the transmission lines occurs,not only the agricultural and industrial production and people’s daily life are affected, but alsopower system safety and stable operation are also imperiled. Therefore, if fault detection andlocation are performed timely and reliably, the fault can be handled early, power restorationcan be assured quickly, and economic loss due to transmission lines fault will be reduced.The present research of fault identification method of transmission lines are expoundedin this paper, the recognition principle and steps are analyzed and compared, and found thesome limitations. For the limitation of each method, a transmission line fault recognitionmethod based on Elman neural network is put forward in this paper. First, through theestablishment of MATLAB fault simulation model, all the types of fault signal are acquired;Then carrying out phase mode transformation of fault current,to getting1mode component,because of the three-phase line electromagnetic coupling phenomena; And then thecomponent is decomposed by LMD to getting the corresponding PF component; Nextcalculating the1(1/2)dimension spectrum entropy of the PF component, to getting the faultfeature vector; At last the feature vectors are input to the Elman neural network trainingsample to achieving fault type identification.The method proposed in this paper is with the analysis of LMD as a treatment method offault signal, this method clearly depicts the fault characteristics of transient signals through1(1/2) dimension spectrum entropy with clear physical meanings. This method Can judge thefault type and fault phase quickly and accurately. Compared with other methods, it has a fasterrecognition speed, more accurate recognition rate, and is not influenced by the transitionresistance, fault location, fault initial angle etc.. Large number of simulation results show thatthe recognition method of the phase selection is feasible and effective in transmission linefault.