Research On The Key Technology Of Dynamic Traffic Path Planning

As the result of the explosion of vehicles’number in China and the further development of urbanization since the1990s, problems such as vehicle congestion, traffic accidents are more and more frequently affect people’s travelling, cause unnecessary losses for the various aspects of social life. There have been some studies focus on dynamic traffic flow planning models and algorithms, but mostly of them are under two extreme conditions of travelers’intentions, lacking traffic flow modeling under the guidance of multi principles and ignoring individual traveler’s tolerance irrationally in the pursuit of the overall system optimization.This subject focus on the above mentioned aspects, proposed a dynamic traffic assignment model DUCSO (dynamic user cost and system optimization) with both personal and global total travel time optimization. In the model, we mainly consider the acceptance degree of the recommended path. Using historical average data as the guarantee of guarantee, dynamic parallel algorithm as an effective addition, this paper put forward an detailed deployment program based on the cloud platform and the Internet of Things. Considering the complexity is too high, this paper puts forward a non-backtracking flow assignment approximation algorithm NBFAA(non-backtracking flow assignment approximation) to solve the model, besides, analyzes the approximate ratio. Facing the uncertainty of traffic flow assignment model, this paper proposes an efficient parallel dynamic path search algorithm as a useful supplement to the model, at the same time, deployed to cloud services platform to obtain performance advantage. Through the simulation, we can see that the application of the model make travelers avoid traffic congestion under the guidance of guidance information, meanwhile, personal travel time is relatively short. And at the same time, the model enhances user’s compliance in the guidance information. The main contributions and innovation points of this paper are:(1) Propose a dynamic traffic assignment model DUCSO with both personal and global total travel time optimization. In this paper, we improved the dynamic model of Wardrop using user’s additional cost as constraint conditions, which makes a qualitative leap in accord with the model and the actual level. Users who accepted the guidance achieve quasi system optimum with little cost.(2) Put forward a feasible path search algorithm with parallel deployment. In this paper, we design a reasonable resolution of parallel ant colony algorithm from the perspective of parallel, which has a realistic significance in the background of vigorous development of the cloud.(3) Present a traffic planning system based on cloud platform and Internet of things, at the same time, design the prototype system. The deployment is based on cloud computing, using Internet of things as the auxiliary means to reduce the amount of calculation and the strain of the cloud platform, and it has a good effect.(4) Propose a traffic network prediction algorithm based on the Internet of things. The algorithm has a small calculation amount, and can be applied to small embedded system with directional characteristics of multicast, at the same time, greatly reduce the processing burden of cloud platform.