Research On Real-time Traffic Flow Guidance For Urban Network Based On System Optimum

The imbalance between traffic demand and the road network supply is the key to many traffic problems.On the condition of the limited improvement of supply side,conducting accurate guidance to traffic demand and equilibrating the distribution of traffic demand in time and space to make the most of the road network resources is considered as an effective method to relieve the dilemma between traffic demand and supply.Since the existing guidance system can only provide simple feedback based on current traffic status and the implement effect is not satisfied.Some researches treat traffic system cost as the only optimization target and did not consider the profit of users,which may result in difficulty in guarantying guidance rate.A dynamic traffic assignment model that consider the cost of traffic system and users' route selection is established in this paper to solve problems above.Also,the guidance plan is studied based on the assignment model,which is considered as the theoretical basis of the real-time traffic flow guidance.The key contents of the paper includes the dynamic OD estimation method for urban road network,the extraction of vehicle routes,and the generation of guidance plans.The research contents and main conclusions are as follows.Obtaining accurate and reliable traffic demand is the premise and basis for the successful implementation of the dynamic traffic assignment process.This paper proposed a dynamic OD estimation method for urban road network that aims to obtain traffic demand matrix of the road network by using observable road sections and turning flow information,which is used as the input of the assignment model.Existing OD estimation models using the dynamic traffic assignment mainly include mathematical analysis methods and simulation methods.All of these models have the disadvantages of high solution complexity,many artificial hypotheses,and large deviations from actual conditions.The deep learning technology is used in this paper to capture the dynamic relationship between traffic demand and traffic counts from the perspective of data and achieved an accurate description of the assignment relationship.Secondly,the paper used heuristic algorithm as the up-level optimization algorithm of the estimation model.Based on the off-line estimation of OD pattern using historical data,the real-time traffic counts are used as the input to quickly calculate the dynamic traffic demand.Using the downtown area of Kunshan City as an example road network,the performance evaluation results show that the proposed model can obtain accurate and reliable dynamic traffic demand in a relatively short time.Meanwhile,compared with the existing commercial software PARAMICS,the proposed model has greater advantages in terms of efficiency and accuracy.On the basis of in-depth analysis of user travel behavior,it is essential to improve the rate of obedience and protect the interests of individual users by incorporating path selection behavior into the dynamic traffic assignment modeling process.Based on real word license plate data and taxi GPS data,a method of trajectory reconstruction for urban roads is proposed in this paper to accurately extract the user's travel path and provide support for the subsequent assignment model.At the same time,in order to estimate the missing link caused by the data quality of the equipment,a dynamic K shortest path search method for urban road network considering the delay of turning and link travel time is presented in this paper,which uses the deleting method as the outer loop and the A* algorithm as the inner loop to quickly searches the dynamic K shortest path.Compared with the genetic algorithm,the proposed method has the advantage of high speed and accuracy.Finally,the paper builds a system optimum based dynamic traffic assignment model is built in this paper to get optimal path selection,and takes this as the objective to study the method of guidance plan generation.Considering the great difficulties caused by a large number of nonlinear constraints in the conventional dynamic traffic assignment model,the paper abstracts the physical road network into a high dimensional space-time network and embeds the constraints of travel time into the road network modeling.The integer programming model is constructed to realize system optimum based dynamic traffic assignment.Further,on the basis of extracting the user's travel path,an effective path set is generated,and the traffic demand is allocated to the user's daily trip path.Finally,aiming at minimizing the difference between the actual guidance effect and the expected effect,considering the influence of the induced device coverage and obedience rate,a method for generating the guidance plan is proposed.The case analysis shows that the proposed method can guarantee the individual user's rights without increasing the total cost of the system.The guidance plan generation method can effectively reduce the differences between the actual guidance effects and the expected effects by adjusting the guidance proportion.