Fault Diagnosis Of Transmission Lines Based On VMD Sample Entropy And KELM

China's power system has been in the leading position in the world through the painstaking research of numerous experts and scholars,and the construction of transmission lines has also developed prosperously.As the highest failure rate part,transmission lines threaten the safe and stable operation of power systems.Short-circuit faults are the most frequent and serious faults in transmission line faults,which may cause system oscillation or even system collapse.Therefore,it is necessary to identify the fault type quickly and accurately when the fault occurs,which is an important prerequisite for the correct action of the relay protection device and the accurate removal of the fault.Fault diagnosis of transmission line is mainly divided into two parts: fault feature extraction and fault type diagnosis.At present,the commonly used feature extraction methods include Fourier decomposition,wavelet transform and empirical mode decomposition.However,the above methods all have some defects: Fourier decomposition can not extract the local feature information of non-linear models;Wavelet transform needs to determine the decomposition scale artificially.Basis function;empirical mode decomposition may lead to the phenomenon of modal aliasing,which will lead to the lack of feature information.This paper combines variational mode decomposition(VMD)with sample entropy to extract fault features of transmission lines.VMD can not only perform fault signal analysis well,but also has good noise robustness.VMD effectively overcomes the shortcomings of the above-mentioned signal analysis methods,and can better realize fault signal processing.Sample entropy is applied to extract fault features.Accurate fault features are obtained.In the construction of fault identification model,the traditional fault identification model is realized by using neural network and support vector machine.Although some achievements have been made,there are still some drawbacks: the convergence speed of neural network is slow,and it is easy to fall into local minimum;it is difficult for support vector machine to construct multi-classifier,which requires a lot of training time to realize multi-type fault recognition.Kernel Extreme Learning Machine(KELM)has higher learning rate and measurement accuracy,better generalization ability,and can effectively overcome the drawbacks of slow learning rate and low diagnostic accuracy of traditional models in Multi-fault type diagnosis.The fault features extracted by VMD and sample entropy are input into the kernel extreme learning machine,which can effectively realize the short-circuit fault diagnosis of power system transmission lines,and the convergence time is short,the diagnosis accuracy is high,and the generalization ability is strong.Finally,an experimental platform is built with MATLAB software to simulate a transmission line in Xinjiang power grid.Several types of faults such as single-phase short-circuit,two-phase short-circuit,two-phase grounding short-circuit and three-phase short-circuit are set up and experimental data are obtained.The proposed VMD and sample entropy method are used to extract features.The obtained experimental data are decomposed by VMD,and the sample entropy of decomposed components is calculated.Then,the extracted features are applied to the diagnosis model of the kernel extreme learning machine proposed in this paper and the traditional model for experimental verification and comparison.The simulation results show that the proposed feature extraction method and diagnosis model have faster diagnosis speed,more ideal rationality,adaptability and higher diagnosis accuracy than the traditional methods and models.