Comparative Study Of Transmission Line Icing Return Period Based On The Various Extreme Value Distributions

Transmission Line Icing accumulated up to a certain limit thickness of which formed ice storm will poses a great threaten to the design and stable operation of the power grid.If they failed to consider the extreme icing conditions during the early construction of Transmission line,which could easily cause the collapse lines,power outages and other major impact.Scientific and rational to estimate the transmission line Extreme Icing is a big trouble that should be addressed for the Power grid planner when they start to routing planning and design criterion.As in the light of actual Light or Repeated icing area,the statistical properties of the Extreme Icing are significant differences,the Extreme Icing of which obeys the extreme value distribution data and Extreme estimation method of application may be different,and the current domestic and foreign scholars have not distinguish the difference but with a common extreme value distribution to estimate a unified analysis,the estimates of which there may exist a comparatively large error.The sites with various icing condition could be fitted to different extreme value distribution.In order to find out the difference,icing accretion data of 4 stations are investigated in this paper.Generalized Extreme Value(GEV)based annual maxima,and Generalized Pareto Distribution(GPD)based Peak over Threshold(POT)and Method of Independent Storm(MIS)is utilized to estimate extreme ice accretion of 50-year return period level.In order to further improve the estimation accuracy,as a limited number of samples under the Extreme conditions,the maximum likelihood estimation method is difficult to eliminate the uncertainty of the results of the problems of parameters estimation.Application of Bayesian estimation method,the unknown parameters of extreme icing value distribution is regarded as a random variable,the icing parameters of GEV distribution and the posterior distribution is estimated by the Monte Carlo method of which based on the adaptive sampling.Then further the extremum icing Estimator.Comparative study of K-S test,?2(chi-square)test and estimated distribution parameters and extreme ice accretion.It suggests that ice former outperform the latter two approaches in the region either heavily or slightly affected by the ice accretion.The extreme ice accretion in regions slightly affected by ice accretion seems follow extreme value distribution of type II while the extreme ice accretion in regions heavily affected by ice accretion follow the distribution of type GPD.Since the extreme value I,II,III distribution can be incorporated in GEV,it seems prudent to estimate extreme ice accretion with GEV.In addition,due to the use of a function that isindependent of the likelihood asymptotic properties of prior information,the Extreme icing posterior distribution obtained by the Bayesian estimation,compare with the result obtained by maximum likelihood method can contain more information;it can express the uncertainty prediction which caused by the parameters uncertainty and cause some improvement on the accuracy of the estimated extremum Icing,the results of a reasonable estimate of overhead transmission line conductor design the allowed Extreme ice thickness has some certain reference.