Research On Modeling And Algorithm Of Urban Commuting Cars Reservation With Road Congestion Condition

As the economy develops rapidly and the city scale expands gradually,there is still no effective traffic management strategy in the face of increasing traffic demand,which makes the traffic congestion increasingly serious.With a view to alleviating traffic congestion,rationally allocating road resources and increasing traffic efficiency,this paper puts forward the concept and scheme of urban road traffic reservation for commuting cars on urban roads on the basis of road flow control.The research contents are as follows:1.Aiming at the problem that the departure time and traffic demand characteristics of commuters cannot be accurately understood,this paper studies the travel plan estimation method of commuting cars based on computational graph.By setting the control function between multi-layer variables and integrating multi-source data,the traffic computational graph is established to estimate the travel plan of commuters,which solves the uncertainty caused by single data source.According to the relationship among OD,path flow,link flow and historical data,the weights of variables in each layer in the computational graph are calibrated by back propagation algorithm.Furthermore,using mobile phone GPS data,floating car data,detector data,etc.to construct a case scenario and estimate the real-time departure time and route of commuters,which can provide data basis for the preparation of commuter travel reservation schedule based on Pareto Improvement.Compared with the single data source prediction method,it is proved that this method can effectively improve the prediction accuracy.2.In order to better reflect the advantages of traffic reservation,the differences among traffic reservation,user equilibrium and system optimal are analyzed by examples,and it is pointed out that traffic reservation is equivalent to real-time system optimization.According to the traffic flow theory and macroscopic fundamental diagram theory,the flow-density curve is divided into free flow,congestion and other states in order to better describe the road state,so as to control the road state more intuitively.For proving the feasibility of traffic reservation,the cumulative flow curve is fitted into a quadratic function form by means of cumulative flow curve tool and differential equation mathematical method,and the existence of Pareto Improvement solution in traffic reservation is proved.Furthermore,the quadratic cumulative flow curve is extended to three times,which proves that Pareto Improvement solution exists in a broad sense in traffic reservation.3.In order to solve the problem of commuters’ travel in traffic reservation,a mixed integer programming(MIP)model based on commuters is established by using the spacetime-state network tool.Based on two scenarios,namely,constant bottleneck outflow rate and declining bottleneck outflow rate in the case of over-congestion,a traffic travel state selection constraint is added to formulate an end-to-end road traffic travel schedule for each commuter.Compared with the traditional traffic flow model based on traffic flow,the characteristics and advantages of this model are demonstrated.In order to better manage the road conditions in real time,this paper puts forward a macroscopic bottleneck flow control method,establishes a flow-based bottleneck flow control model,and solves it by a three-dimensional dynamic programming algorithm based on cumulative flow curve.With the help of DTALite open-source traffic assignment software,it simulates the actual case of West Third Ring Road in Beijing.By comparing the road conditions before and after the reservation of major bottlenecks,it shows that the macroscopic flow control model has a good flow control effect.4.In view of the problem that commercial software is slow in solving large-scale road network and heuristic algorithm can’t get the optimal solution,according to the characteristics of decision variables in MIP model,the alternating direction multiplier method(ADMM)is used to dualize,decompose and linearize the micro-MIP model based on commuters.To facilitate the calculation and get the optimal solution faster,the original planning problem is regarded as a commodity flow problem,and it is dually decomposed into several subproblems,that is,the shortest space-time problem for each commuter.By analyzing the case of the large-scale road network in the West Third Ring Road of Beijing,according to the comparison of road traffic conditions before and after the reservation and commuter’s travel utility,the efficiency of traffic reservation in congestion control is verified.