Research On Adaptive Variable Approach Lane In Ground Connection Section Of The Elevated Highway Exit Based On Promoted MACA Model

With the development of urbanization and the increase of vehicle ownership in China,the ground traffic network is difficult to carry huge traffic flow.As a threedimensional and interactive traffic diversion facility,elevated highway has become an important means to solve traffic congestion and improve traffic efficiency in many cities.Due to the limited land space available in the city,and in order to maximize the utilization rate of the elevated expressway,the elevated exit ramp is often set near the ground intersection.The vehicles on the lower ramp with different steering intentions and the vehicles on the ground auxiliary road will form a complex traffic interweaving area in the ground connection section in the process of selecting the inlet of the intersection,which affects the traffic efficiency of the road section and generates traffic safety risks.Therefore,the vehicle operation law of the ground connection section of the elevated exit is analyzed in depth,and the key factors affecting the vehicle driving decision are found,and the traffic efficiency and safety evaluation index system of the ground connection section of the elevated exit is constructed.It has become an urgent issue to explore the effective means of traffic control in the ground connection section of elevated export at this stage.To this end,this paper relies on the National Natural Science Foundation of China(Grant No.51778141)and the Jiangsu Province Graduate Research and Practice Innovation Program Project Study on Traffic Conflict Prediction and Control Scheme Evaluation of Urban Elevated Exports(KYCX20＿0138),and takes the off-ramp terminal exit of the south and east expressway of Jiangbei Avenue Expressway in Nanjing as the research case,and the study was structured in following five aspects:Firstly,the UAV(unmanned aerial vehicle)was used to collect the traffic flow video of the off-ramp terminal exit section.A traffic flow video recognition algorithm framework based on YOLOv4 vehicle detection and SORT vehicle tracking was proposed to realize high precision and high efficiency vehicle original trajectory extraction.At the same time,the data processing and reconstruction methods of vehicle original trajectory were proposed from the aspects of region of interest selection,pixel coordinate transformation,trajectory data smoothing and broken trajectory matching,and a complete and effective vehicle trajectory data set was established.On this basis,the macroscopic traffic flow characteristics of the ground connection section of the elevated exit were described and analyzed from four aspects: vehicle type characteristics,vehicle travel time characteristics,vehicle initial speed characteristics and vehicle inlet selection characteristics.Secondly,this study extracted the data information of car-following and lane changing data from the vehicle trajectory data set,and analyzed the microscopic driving behavior characteristics of vehicles in the off-ramp terminal exit section.For carfollowing,this paper selected four classical car-following models,Gipps model,GHR model,OV model and IDM model,and calibrated model parameters based on carfollowing data under normal driving state,deceleration and parking state and start acceleration state,explores the methods for improving car-following model,and establishes a phased car-following prediction model.For lane-changing behaviour,a hybrid Logit vehicle lane change prediction model considering driver heterogeneity was constructed for descretionary lane-changing and mandatory lane-changing vehicles in the off-ramp terminal exit section,and the range of model parameters was given according to the sensitivity of different drivers to external stimuli.Thirdly,a promoted multi-agent cellular automata(MACA)traffic flow simulation method was introduced to complete the construction of the simulation model of the offramp terminal exit section,including the information control layer and the vehicle layer,respectively.Based on the driving behaviour(car-following and lane-changing model)prediction model of the off-ramp terminal exit section,the traditional Na Sch and STNS cellular evolution rules were improved to achieve the high-precision simulation of the traffic flow of the off-ramp terminal exit section.Fourthly,based on the traffic flow simulation data of the off-ramp terminal exit section from the improved MACA model,the short-term and long-term memory network(LSTM)short-term traffic flow prediction model based on Bayesian optimization was established to realize the prediction of different steering traffic volume and total traffic volume of each lane input in the future signal cycle.On this basis,the traffic pressure balance index of the import lane was proposed as the traffic efficiency index.At the same time,the traffic conflict technology was introduced to evaluate the traffic safety of the ground connection section of the elevated exit,and the selection method of the alternative index threshold of traffic conflict was discussed.The generalized linear model was constructed to establish the relationship between the number of traffic conflict risk and the predicted value of short-term traffic volume.Fifthly,an adaptive variable approach lane control system is constructed,which considered the traffic pressure balance index of the entrance lane and the number of traffic conflict risks in the predicted signal cycle.The layout scheme of the adaptive variable approach lane is designed from three aspects : the buffer and pre-signal design of the approach lane,the dynamic LED indicator design of the lane steering,the vehicle detector and the roadside control unit design.The adaptive variable approach lane operation control scheme was designed from two aspects : buffer security control scheme and adaptive variable approach lane conversion timing judgment.Finally,the effectiveness of the control system is verified by simulation examples.This paper provides a method for the collection and analysis of traffic data in the ground connection section of the elevated exit,reveals the internal motivation of microscopic driving behavior decision-making of vehicles in the ground connection section of the elevated exit,and improves the traffic operation evaluation system of the ground connection section of the elevated exit from the perspective of efficiency and safety,which provides theoretical support for the layout and operation control of variable steering lanes.