Instantaneous And Permanent Fault Identification Based On VMD And SVM

When a fault occurs on the transmission line,the single-phase automatic recloser does not determine the nature of the fault,and it starts to reclose quickly.When a transient fault occurs on the line,the auto-reclosing fault disappears after the reclosing,ensuring the reliability and continuity of the transmission line power supply.However,when the auto-reclosure coincides with a permanent fault,it will cause a secondary impact on the power grid,causing damage to the power system and power equipment.In this paper,based on the study of the voltage of the fault phase terminal when the single-phase ground fault occurs on the transmission line,the method of extracting the transmission line fault feature based on VMD information entropy is designed,and the two methods based on VMD information entropy-BP neural network and VMD information entropy-SVM are studied.A fault property recognition method,by comparison,found that VMD information entropy-SVM fault property recognition method has better noise resistance and real-time performance.First of all,this paper analyzes the voltage characteristics of the disconnected phase between the transient and permanent faults of the transmission line with parallel reactors from a theoretical perspective,combined with the characteristics of the first and second arcs,and built 500kV in the ATP/EMTP simulation software The simulation model of the transmission line with parallel reactors simulates the voltage waveform at both ends of the fault phase at different voltage phase angle differences,different transition resistances,and different fault locations,to obtain the simulation data of the phase terminal voltage of the single-phase grounding fault of the transmission line.The results show that the transient voltage waveforms of instantaneous faults and permanent faults in single-phase ground faults are obviously different.Secondly,from the perspective of signal analysis,variational mode decomposition(VMD)is performed on the voltage waveform of the line terminal that has a single-phase ground fault and contains a complex nonlinear signal,and all the modes of signal decomposition are obtained by iterative update Weight.The decomposed modes include the modes of the main signal and the modes of noise.Reconstructing the modes of the main signal achieves the effect of filtering harmonics and denoising.After decomposition,four intrinsic mode functions are obtained Intrinsic Mode Function(IMF),the information entropy is calculated for each modal component to form a set of four-dimensional feature vectors.Finally,BP neural network and Support Vector Machine(SVM)are introduced to study and compare the identification method of single-phase ground fault nature of transmission line between VMD information entropy-BP neural network and VMD information entropy-SVM.Simulation results show that,without Gaussian white noise,both VMD information entropy-BP neural network and VMD information entropy-SVM can accurately distinguish between transient faults and permanent faults.After adding the Gaussian white noise,the accuracy of the VMD information entropy-BP neural network in distinguishing the nature of the faults dropped to 97.5%and 98.75%,respectively.VMD information entropy-SVM can still accurately distinguish the nature of the fault,and this method has little effect on the discrimination results when changing the transition resistance,fault location,and power phase angle difference.