Research On Fault Location Of UHVDC Transmission Based On MPGA-Elman Network

Ultra-high voltage direct currentin China is in full swing and is a very important link in transmission system.Because the geographical location of the transmission line is very complex and the transmission distance is relatively far,the ability to detect and eliminate the fault accurately and accurately is very important for the reliability of the transmission line when there is a fault,but when the transmission line has a fault,Relying solely on the inspector to find the fault point is not only inefficient,but also transient faults such as flashover often cause damage to the line,usually without obvious signs of damage,and greatly increase the difficulty of manual line patrol.Therefore,this paper proposes to use MPGA-Elman network to align directly.The current transmission line carries on the distance measurement.Taking Yunguang UHVDC system as the research object,the UHVDC transmission model is established.Because the fault traveling wave will occur in UHVDC transmission line,there is a certain correlation between the natural main frequency and the fault distance,but the inherent main frequency is greatly affected by the line parameters and the extraction accuracy is poor.However,the spectrum energy characteristics near the inherent main frequency are obvious,and when the fault resistance is different and the fault distance is different,the spectrum energy is different.Based on the above reasons,the fault voltage in the case of short circuit on the positive circuit of the simulation model is extracted,and then the Karenbauer matrix is used.De-coupling.After decoupling,the 1-mode component is decomposed and reconstructed by wavelet packet,and the energy of each frequency band is calculated.In view of the fact that Elman neural network can establish a fuzzy relationship between fault distance and spectrum energy without exact function relation,this paper uses Elman to locate DC transmission lines.Because the spectrum energy values of each frequency band vary greatly,which is not conducive to the convergence of Elman networks,the ratio of energy ratio to the total energy value of each frequency band is introduced as the input of the network.Because of the structural limitation,the convergence speed of Elman neural network is very slow and it is easy to converge to the local minimum.Therefore,the multi-population genetic algorithm(MPGA)is used to optimize the weight and threshold of Elman network and speed up the Elman.The convergence speed improves the timeliness to a certain extent.The energy ratio is sent to MPGA-Elman network as fault characteristic quantity,and the trained MPGA-Elman network can be used for fault location.The simulation results show that the fault location error based on MPGA-Elman neuralnetwork can be achieved under different transition resistance and fault distance.Within,the convergence speed is much higher than that of Elman and GA-Elman networks,and the real-time performance is better,so it is suitable for fault location.There are 42 graphs,7 tables and 56 references in this paper.