| Because the increasing voltage level keeps increasing the height of the t ower,and the transmission capacity of the line is also increasing,which leads to the increase of the line corridor,the probability of high-voltage transmis sion lines being subjected to lightning strikes is increasing.Lightning strikes on transmission lines can have a severe impact on the power supply system.The research and analysis of lightning strikes and short-circuit fault informati on is the basic premise for dealing with data system failures and strengthening lightning protection measures.Identifying lightning strikes and short-circuit faults on transmission lines can ensure targeted on-site troubleshooting,which is of great significance for rapid troubleshooting.The article also considers the segmented wave impedance model of the tower through the ATP-EMTP simulation system,and conducts lightning strike simulation of 220 kV transmission line.Four types of traveling wave waveforms of common power frequency short-circuit fault,non-fault lightning strike,counterattack fault and bypass fault are obtained.The simulation results show that different types of lightning current signals have different waveform characteristics in the time-frequency domain,and the mechanism of overvoltageg eneration is analyzed,which lays a solid theoretical foundation for feature an alysis.Existing transmission line lightning strike identification technology methods are mostly distributed and classified to analyze the time or frequency do main characteristic information of the signal,and there may be insufficient a nalysis capabilities for the comprehensive identification characteristics of the lightning current in the time and frequency domain.In this paper,three eigen vectors of average zero-crossing rate,current integral eigenvalue and approxi mate entropy value are determined in the time domain scale,and the wavelet energy ratio and teager energy spectrum feature quantities are extracted based on the frequency domain scale and teager energy operator,respectively.The distinction between non-lightning strike fault and lightning strike fault,lightning strike disturbance and lightning strike fault,and counterattack fault is realized.It is obtained from simulation and simulation.Under the proposed criteria,the identification task can be completed by different interference conditions and identification factors.A single criterion is prone to misjudgment when it is recognized,and multiple criteria are easy to be judged because of the high dimension.In order to overcome the above problems,the article based on the analysis and research of the support vector machine algorithm,A method of intelligent recognition system of lightning strike fault based on support vector machine is proposed and verified.After completing the binary classification of the recognition process and the distribution of recognition vectors,a support vector machine algorithm model was built to realize an intelligent model that can recognize various fault types.A large number of simulation experiments show that the proposed method has higher accuracy and better recognition performance.Aiming at the problems that the average zero-crossing rate and the results of traditional traveling wave head detection algorithms are greatly affected by noise,this paper proposes a noisy traveling wave detection algorithm based on multiple variational mode decomposition.The simulation results show that even under severe noise interference,the algorithm can still distinguish between short-circuit faults and lightning strike signals,and at the same time detect the fault traveling wave head characteristics more accurately.Finally,the real data is identified to verify the reliability of the method. |
PDF Full Text Request