Research On Fault Identification Method Of Transmission Line Based On IKH Optimizing S-GRU

Aiming at the serious harm of short-circuit faults of transmission lines and the low fault identification rate,a transmission line fault identification method combining variational modal decomposition(VMD)-permutation entropy(PE)and the siamese gated recurrent unit(S-GRU)neural network optimizated by improved krill herd(IKH)algorithm is proposed.It is mainly divided into two aspects: fault feature extraction and fault type identification.In view of the problem that the modal decomposition level K of the VMD algorithm has no clear basis for selection,the VMD parameter is optimized by the instantaneous frequency average,and the decomposition level K is determined.The fault voltage signal of transmission line is decomposed by VMD,and the permutation entropy of each component after decomposition is calculated as the fault eigenvector to avoid mode aliasing.Since the krill herd(KH)algorithm will cause the problem to fall into a local optimum,the perturbation operator ? can be introduced to improve the optimization performance of the KH.The improved IKH algorithm is used to optimize the GRU neural network fault identification model to strengthen the generalization ability of GRU.In order to achieve better identification effect when there are fewer training samples,propose to build the S-GRU fault identification model by the siamese neural network(SNN)to ensure the best fault identification effect when there are fewer fault samples.The fault feature vector is input into the trained fault identification model,and the similarity of the two input samples is measured to determine the short-circuit fault type of the transmission line.The transmission line simulation model is built by MATLAB and four typical short-circuit faults(Ag,AB,ABg,ABC)are set up respectively.The short-circuit fault identification is performed at different fault phase angles,different transition resistances and different fault locations respectively.The experimental results show that the average identification accuracy reaches more than 96%,which proves the effectiveness and applicability of the method.Compared with the traditional fault identification method,it finally proves the superiority of this method.This thesis has 47 figures,11 tables and 70 references.