Latest revision as of 21:22, 3 February 2021

Abstract

In this paper, considering the usage of self-driving, I research the congestion problems of traffic networks from both macro and micro levels. Firstly, the macroscopic mathematical model is established using the Greenshields function, analytic hierarchy process and Monte Carlo simulation, where the congestion level is divided into five levels according to the average vehicle speed. The roads with an obvious congestion situation is investigated mainly and the traffic flow and topology of the roads are analyzed firstly. By processing the data, I propose a traffic congestion model. In the model, I assume that half of the non-self-driving cars only take the shortest route and the other half can choose the path randomly. While self-driving cars can obtain vehicle density data of each road and choose the path more reasonable. When the path traffic density exceeds specific value, it cannot be selected. To overcome the dimensional differences of data, I rate the paths by BORDA sorting. The Monte Carlo simulation of Cellular Automaton is used to obtain the negative feedback information of the density of the traffic network, where the vehicles are added into the road network one by one. I then analyze the influence of negative feedback information on path selection of intelligent cars. The conclusion is that the increase of the proportion of intelligent vehicles will make the road load more balanced, and the self-driving cars can avoid the peak and reduce the degree of road congestion. Combined with other models, the optimal self-driving ratio is about sixty-two percent. From the microscopic aspect, by using the single-lane traffic NS rule, another model is established to analyze the road Partition scheme. The self-driving traffic is more intelligent, and their cooperation can reduce the random deceleration probability. By the model, I get the different self-driving ratio of space-time distribution. I also simulate the case of making a lane separately for self-driving, compared to the former model. It is concluded that a single lane is more efficient in a certain interval. However, it is not recommended to offer a lane separately. However, the self-driving also faces the problem of hacker attacks and greater damage after fault. So, when self-driving ratio is higher than a certain value, the increase of traffic flow rate is small. In this article, that value is discussed, and the optimal proportion is determined. Finally, I give a nontechnical explanation of the problem.In this paper, considering the usage of self-driving, I research the congestion problems of traffic networks from both macro and micro levels. Firstly, the macroscopic mathematical model is established using the Greenshields function, analytic hierarchy process and Monte Carlo simulation, where the congestion level is divided into five levels according to the average vehicle speed. The roads with an obvious congestion situation is investigated mainly and the traffic flow and topology of the roads are analyzed firstly. By processing the data, I propose a traffic congestion model. In the model, I assume that half of the non-self-driving cars only take the shortest route and the other half can choose the path randomly. While self-driving cars can obtain vehicle density data of each road and choose the path more reasonable. When the path traffic density exceeds specific value, it cannot be selected. To overcome the dimensional differences of data, I rate the paths by BORDA sorting. The Monte Carlo simulation ...

Original document

The different versions of the original document can be found in:

• https://aip.scitation.org/doi/pdf/10.1063/1.5033806

• http://aip.scitation.org/doi/pdf/10.1063/1.5033806,

http://dx.doi.org/10.1063/1.5033806

• https://aip.scitation.org/doi/abs/10.1063/1.5033806,

https://ui.adsabs.harvard.edu/abs/2018AIPC.1955d0142J/abstract,

https://academic.microsoft.com/#/detail/2799654050