Study On Traffic State Identification Of Expressway In Complex Environment Of Mountainous City Based On FCM

Expressway links large areas of city and shoulders a large size of half and long distance traffic volume,which is the main artery of urban traffic with the advantages of smooth,fast and comfortable.With the rapid development of urbanization and increase in vehicle ownership,these characteristics of expressway are fading gradually now.Different from the plain city,mountain city expressway terrain environment is complex,expressway traffic congestion intensity is greater when they have same traffic volume.In order to alleviate the traffic pressure of fast track,it is significant to study the method for identifying the traffic state of expressway in complex environment of mountainous city.The main topics are addressed in this paper as followings:Firstly,this paper summarizes the research status about identified algorithm of traffic state domestic and foreign,analyzes the problems in proposed algorithm models,and puts forward Fuzzy C-Means(FCM)with mature theory and well pattern recognition ability as the basis model for identification of traffic state.Secondly,analysis of traffic characteristics.It proposes the layered and the free road network layout of mountainous city,puts forward the inconsistent road standards by analysis of examples;it presents the complexity of traffic operation in mountainous expressway qualitatively and quantitatively from road slope,interchange spacing,road junction three aspects;it also proposes that many fixed bottlenecks from bridge and tunnel,poor connectivity of road network are both the special reasons of traffic congestion in mountainous expressway.At last,this paper points out the identified complexity of traffic state due to the instability of traffic flow and the obvious difference of traffic state threshold.Thirdly,study on traffic state classification.Through the analysis of domestic and foreign existing traffic state classification method,parameters and classification standard of different traffic state,combining with the common traffic states partition and actual traffic flow characteristics determines smooth,less smooth,slow,congestion as the paper's traffic states.Fourth,study on the parameters of traffic state identification.It analyses the influence of different traffic parameters on identifying traffic state,summarizes common used traffic parameters and its application status,determines the traffic parameters should be selected from the volume,speed and occupancy in this paper.The optimal parameter combination is determined according to the highest accuracy,through comparing the idenfied state of different parameter combination with the actual traffic state.The results show that volume and velocity is the best parameter combination.Fifth,the weighted index m is researched.This paper analyzes the influence of the weighted index on traffic state identification,puts forward that the optimal value of m is made by clustering accuracy,distance between classes,inter-class distance,value of the objective function.When the higher of clustering accuracy,the smaller of inter-class distance,the larger of distance between classes,the smaller of objective function value are both coming out,the perfomance of state identification will be best.Studies have shown that the optimal value of m is 2.25.Sixth,study on traffic state identification.It proposes the traffic state of singlelane or multi-lane is determined according to the maximum degree of membership;to identify traffic state of road unit,it is obtained by the state value of adjacent section and the average travel speed of road unit.The results show the identified accuracy of multi-lane traffic can reach 96.67% with a certain sample size,and the above methods are feasible.The online states of single-lane and multi-lane would be identified by the the maximum principle of euclidean distance at the same traffic state,which is between simples and typical clustering centers;and the traffic state of road unit is identified through state of adjacent section and the average travel speed.The results show that the accuracy of online traffic state identification based on multi-lane can reach 86.11%,and road unit is up to 83.33%.Based on the comprehensive research,this paper establishes a complete framework of traffic state identification in complex expressway of mountainous city based on FCM algorithm,and provides the theoretical guidance for predicting traffic state in expressway in mountainous city.Finally,the direction of further studies is discussed.