Research On Methods For Traffic State Identification Based On SAGA-fCM

The continuous growth of urban road traffic flow leads to the deterioration of traffic operation,which leads to more and more frequent traffic congestion.How to accurately and real-time identify traffic conditions has become an important research topic.With the help of real-time and effective traffic state identification technology,traffic managers can fully grasp the traffic status of urban roads,by timely publishing the current traffic situation through relevant platforms,traffic congestion can be avoided to a large extent,so as to achieve real-time traffic guidance and improve the efficiency of the entire road traffic network.Firstly,the characteristics of commonly used traffic flow parameters are studied.Average traffic volume,time average speed and time occupancy are selected as the parameters to distinguish urban road traffic status.By comparing the time-varying images of urban road traffic flow at different time intervals,a 5-minute time interval is selected to distinguish traffic status.Secondly,this paper introduces the principle of Fuzzy C-Mean Clustering(FCM)algorithm,and proposes to use cluster validity function to determine the optimal number of classifications in the FCM algorithm.Then,in view of the randomness of FCM algorithm in selecting initial clustering centers,which makes the results easily fall into local optimum solution and makes the algorithm less stable,this paper improves the traditional fuzzy C-means clustering algorithm(FCM),and proposes an improved FCM clustering algorithm for urban road traffic status based on genetic simulated annealing algorithm(SAGA).The experimental results show that the proposed SAGA-FCM algorithm can not only effectively overcome the problems existing in the initial clustering center selection process of FCM algorithm,but also has faster convergence speed and better stability compared with FCM algorithm.Finally,this paper introduces what is artificial neural network,and introduces the principle of BP neural network and radial basis function neural network.Then the GRNN and PNN neural networks which also use radial basis function as pattern classification are introduced.Then,this paper extracts the characteristics of various traffic state data sets clustered by SAGA-FCM algorithm,uses GRNN and PNN algorithms to learn different traffic state data sets,and establishes a decision-making model of urban road traffic state discrimination based on GRNN and PNN.Examples show that the average correct rate of GRNN and PNN models after 10 runs is above 95% and the average running time is less than 1 second in the traffic state discrimination method based on SAGA-FCM,GRNN and PNN proposed in this paper.This proves that it is feasible and reasonable to use GRNN and PNN models as traffic state discriminant models.