Research On Identification Of Transmission Line Fault Based On Deep Learning

With the rapid development of our country,transmission lines are spread all over the country.Because the transmission lines are exposed outdoors and are affected by environmental factors,they are prone to failures.However,the cause of the fault is usually not easy to identify because the location of the transmission line is complex,it is difficult for personnel to patrol the line,it takes a long time,and it causes a lot of inconvenience to the user.Therefore,the research on the identification of the causes of transmission line faults is of great significance nowadays.This paper studies six types of transmission line faults,namely: lightning strikes,cranes hitting lines,bird flashes,pollution flashes,tree flashes,and wildfires.Through in-depth research,it is found that the fault recorder data and environmental factors have an important correlation with the cause of the fault,so the fault characteristics are extracted from the fault recorder and environment,and then the fault is identified.Because the correlation between fault characteristics and fault causes is very complicated,this article uses three algorithms,namely support vector machine,BP neural network,and deep learning to identify.The inputs of the three algorithms are the time of occurrence of the faulty line,the season of occurrence,weather conditions,the resistance of the transition resistance,the harmonic components of the zero sequence current and the zero sequence voltage,and the distortion of the zero-crossing point of the fault phase current.The output of the three algorithms is The category of the cause of the failure.In order to compare the identification effects of the three algorithms,this paper established a support vector machine algorithm model,a BP neural network algorithm model,and a deep learning algorithm model.A large amount of data from a certain province was put into it for training and testing,and the deep learning algorithm was obtained through comparison.The recognition rate of failure causes is the highest,and after DS evidence fusion is adopted,the recognition rate of deep learning is more stable.