Research On Lane-level Traffic Flow Prediction And Optimization Control On Urban Arterial Roads

With the growth of urban transportation demand and the increasing number of motor vehicles in China,the contradiction between the existing road resources and travel demand is becoming increasingly prominent.Urban arterials carry the main traffic flow of urban travel.The prediction of lane-level traffic flow at intersections on urban arterials is crucial for the formulation of reasonable traffic management measures,and the coordinated control between intersections on urban arterials is of great significance for alleviating urban congestion and improving travel efficiency.This dissertation mainly studies the prediction method of traffic flow at the lane-level of urban arterials and the optimization control method of urban arterials.The main contributions of this dissertation are as follows:(1)Research on prediction of lane-level traffic flow on urban arterialsTwo short-term traffic flow prediction models for two different scenarios are established based on the actual traffic flow data on urban arterials.A lane-level traffic flow prediction model based on Catboost is built by analyzing the spatiotemporal correlation of traffic flow,conducting feature engineering,feature selection,and model optimization.A lane-level traffic flow prediction method called lane-GAGRU,which combines graph attention network and gated recurrent unit,is proposed for the road network traffic flow data.The effectiveness and superiority of the models are validated through experiments using real lane-level traffic flow data in Qingdao city.(2)Research on optimization control method of traffic flow on urban arterialsA comprehensive optimization model for urban arterials named PAM-BAND is proposed,which integrates phase optimization and queue dissipation effects and incorporates lane-level traffic flow prediction data for predictive control.In response to the limitations of existing models,split phase is introduced to broaden phase selection,establish the relationship between intersection queue dissipation time and phase difference,and obtain the optimized control model PAM-BAND.The lagging effect of traditional schemes is eliminated by incorporating lane-level traffic flow prediction results into the optimization model.Experimental results show that the PAM-BAND model improves the indicators such as green wave bandwidth,average delay,average number of stops,and average travel time compared to the traditional green wave control model,and the control effect is further improved by introducing lane-level traffic flow prediction.(3)Design and implementation of intelligent traffic management and control systemAn intelligent traffic management and control system for urban arterial roads is designed and implemented.The two core requirements of an intelligent and efficient system are analyzed first,and the system architecture and functions are designed accordingly.The functions and logic of the intelligent analysis module and arterial coordination control module are described in detail.Finally,the various modules are implemented,and the intelligent traffic management and control system for urban arterial roads is built.