Research On Arc Characteristics And Diagnosis Method Of Series Fault In Low Voltagege Power Distribution System

Series arc fault is a key cause of low voltage distribution system electrical fire.The arc current is limited by the load,especially when shielding load(Electric drill,electromagnetic oven)is in existence,the fault arc characteristics are covered,which can not be effectively protected by the traditional protection devices.It is urgent to carry out the research on the characteristics analysis and diagnostic methods of series arc fault.In order to realize accurate diagnosis of low voltage series arc fault,firstly,a carbon electrode contact arc fault experimental platform was built according to the UL1699 and GB14287.4 standards.The series arc fault experiments are carried out by selecting the representative linear and nonlinear loads,which provides a reliable data source for the following research.Secondly,the optimal delay time,the best embedding dimension were calculated by using C-C algorithm.At the same time,the high phase space was rebuild,and the arc voltage and current signal which put into the high dimensional space were analysised.The maximum Lyapunov exponent of arc voltage and current signal was calculated by using the small-data method.In addition,the chaotic characteristics of the low voltage air arc voltage and current signal were analyzed by selecting the distribution law of the phase plane and the maximum Lyapunov exponent as the indicator.The results show that the phase locus of arc voltage was clustered and Arc current signal has fractal characteristics due to the arc has chaotic characteristics.It provides the basis for the design of arc fault diagnosis method.Finally,aiming at clustered distribution characteristics showed by phase track of fault arc voltage signal,the arc fault diagnosis method based on K-means clustering algorithm was designed.Experience proved that the accuracy is higher than 98%,which can be used to detect the series arc fault in a particular location or a particular line segment.According to the fractal characteristics showed by orbit of arc fault current signal,a new arc fault diagnosis method with multi features fusion was proposed using volume dimension and correlation dimension respectively to extract characteristic vector of arcing fault,and using the Least Squares Support Vector Machine(LSSVM)to construct arc fault diagnosis device.The method is more accurate than 95% and can not be limited by arc position,which can be used for low voltage series arc fault diagnosis.The research results not only enrich the fault characteristics of arc voltage and current signal,but also realize the accurate diagnosis of series arc fault in low voltage circuit.It can provide theoretical guidance for design of arc model and electric fire monitoring detector based on the chaotic characteristics of fault arc voltage and current.