Research On Traffic State Discrimination Of Expressway Considering The Influence Of Truck

With the continuous development of highway transportation in our country and the continuous improvement of freight demand,the proportion of trucks in the composition of expressway vehicles is increasing.Due to the certain performance differences between trucks and cars,in the process of mixed passenger and cargo,the interference between vehicles is serious,which is easy to cause traffic safety accidents,which increases the difficulty of highway management and control.In order to manage and control the mixed passenger and freight highway more effectively,this thesis puts forward a traffic state discrimination method considering the influence of trucks,and constructs the highway traffic state discrimination and decision-making analysis model.First of all,by analyzing the expressway operation characteristics and traffic flow parameters,this thesis selects the flow,speed and time share as the discriminant indexes of expressway traffic state,and according to the existing traffic state division standard,a relative division standard of traffic state based on fuzzy clustering is proposed.Secondly,in order to judge the traffic state of expressway more accurately,this thesis analyzes the influence of truck proportion on traffic flow parameters under different flow through VISSIM simulation from the point of view of moving bottleneck,driver and traffic flow state characteristics.Combined with the results of questionnaire survey,this thesis deeply probes into the influence of truck on driver,and takes the truck proportion of truck influence factors as an important index to distinguish traffic state.Then,considering the influence of trucks,the independent weight method is used to calculate the weights of four discriminant indexes: flow,speed,time share and truck proportion.then genetic simulated annealing algorithm is introduced to solve the random problem of initial clustering center selection,and a highway traffic state discrimination model based on weighted genetic simulated annealing algorithm and fuzzy C-means clustering algorithm is constructed.Finally,according to the four kinds of traffic state data sets divided by the discriminant model constructed in this thesis,the expressway traffic state discriminant decision analysis model based on Light GBM is constructed.Combined with the actual road data,the feasibility and effectiveness of the SAGA-FCM discriminant model and Light GBM discriminant decision analysis model based on weighted characteristic parameters are verified and analyzed.The results show that compared with the traditional FCM discriminant model,when the number of iterations is about 20,the objective function of the improved SAGA-FCM discriminant model converges to the minimum,and the convergence speed is faster,which makes the result of traffic state discrimination more reasonable.Compared with the common RF,KNN and GBDT models,the performance evaluation index values of Accuracy,Precision and Recall of Light GBM decision analysis model are more than 0.9860,which is better than other models,and runs faster,which verifies its accuracy and effectiveness.To a certain extent,the research results of this thesis can provide scientific theoretical support for highway traffic management departments to carry out traffic management and control.