Research On Traffic Flow Prediction And Control Method For Main Bottleneck Section Of Freeway

The bottleneck section of freeway often appear at fixed locations such as bridge junction,lane narrowing and junction area,or any position of the main line affected by construction and accidents.Therefore,how to define,classify and accurately identify the bottleneck sections and the difficult problems of traffic flow prediction and effective contro,have been the key and difficult points in the field of operation management.The study of those problems is of great theoretical and practical significance.This paper takes the bottleneck section of the mainline freeway with narrowing lanes as the main research object,and aims to improve its capacity by studying key technologies such as traffic flow prediction,variable speed limit control and control effectiveness evaluation of the bottleneck section.The specific content is summarized as follows:(1)This paper expounds on the definition of bottleneck sections and divided bottleneck sections into different types from the causes,formation speed,and occurrence frequency.On this basis,the internal causes and evolution characteristics of different types of bottleneck sections were deeply analyzed,and the common traffic control strategies of bottleneck sections were summarized,which laid a foundation for further research in this paper.(2)To reveal the change rule of traffic flow running state in bottleneck sections,two-lane and three-lane cellular automata models considering single-lane reduction were constructed from the perspective of microsimulation.The spatiotemporal distribution characteristics of traffic flow on two-lane and three-lane freeways under different control conditions were simulated and analyzed respectively.The simulation results showed that lane reduction results in uneven distribution of traffic volume in each lane and a large dispersion degree of vehicle speed,leading to the formation of a bottleneck in the section.By applying a certain speed limit control,the balance of traffic flow could be improved and the generation of bottlenecks could be delayed or avoided.(3)To accurately identify bottleneck sections,a traffic state division method based on K-means++and a traffic state prediction method based on Long-Short Term Memory networks(LSTM)is proposed.Firstly,through the processing of ETC door frame data,the similarity and stability of traffic flow and speed are analyzed,and traffic states are divided into different categories based on the K-means++algorithm.The classification results show that the traffic flows of the three different traffic states differ roughly by 4vehicles,with a 10s difference in travel time and a standard deviation of about 1.On this basis,the spatial similarity and temporal continuity of different sections and different traffic state categories were fully considered,and the traffic state prediction model based on LSTM was constructed.By analyzing the impact of different variables on predictive accuracy and considering the research context of this paper along with the characteristics of ETC gantry data,a set of variable combinations that are most suitable for model prediction in this scenario has been chosen in this study:closed-loop online learning duration(T_s)=1 min,time step(T_r)=5 min,open-loop prediction duration(T_p)=1 min.This selection has effectively enabled the accurate classification of traffic flow states for a future time interval.It should be noted that while this combination has demonstrated favorable outcomes in the experiments conducted within this paper,it does not imply universal applicability.The optimal combination might vary based on factors such as geographical location,road type,and prevailing traffic conditions.(4)To make speed limiting strategy scientifically,a variable speed limiting control model based on situation prediction was proposed.Firstly,the variable speed limit system architecture for the bottleneck section of the main freeway was constructed,and the information interaction process of the system was defined.Then it put forward the design rules of variable speed limits under traffic congestion and adverse weather(foggy day,rainy day),and the concrete quantitative method of variable speed limit rules.Finally,by building a simulation environment,the suggested value of early warning speed under different bottleneck road scenarios and different traffic demands was given.The simulation results showed that the variable speed limit system would decrease the traffic efficiency and increase the speed fluctuation at low flow rate(600-800 pcu/h/lane).Under medium flow(900(pcu/h/lane)),reasonable speed limit could inhibit the time and degree of traffic congestion at bottleneck points.Under high traffic flow(1000-1100(pcu/h/lane)),variable speed limit had limited effect on alleviating traffic congestion and improving traffic efficiency,but reasonable reduction was expected to reduce speed deviation and improve traffic safety.(5)To effectively evaluate the implementation effect of control measures on bottleneck sections,the paper constructed an evaluation index system of control measures on bottleneck sections of freeways from four aspects:traffic safety,service level,economic benefits,and road traffic facilities.As well as the measurement methods and classification standards of each index,a quantitative evaluation model of the effectiveness of control measures based on the analytic hierarchy process was established.By taking a freeway as an example,the validity of the constructed evaluation method system was verified.