Research On The Application Of Genetic Neural Network In Traffic Control

From the first traffic light appeared in the intersection at grade of London, the study of traffic control has lasted for about 140 years. In 1963, the first centralized traffic signal control system was completed in Toronto, which symbolized that traffic control entered into the times of systematization. At present, some urban traffic problems such as urban traffic congestion, and the imbalance between the development of traffic and the usage of resource, environmental pollution caused by transportation etc. are restricting economic development in almost every nation. How to obey the principles with regard to environmental capacity, fair, harmony as well as the rate of resources wearing and make good use of ever appearing advanced science and technology into traffic control to resolve urban traffic problems will stay a focal point of traffic control.Thanks to the extensive authorization of ITS by the world from 1994, ATCS as an important component of ITS had entered another time. By means of advanced intelligent control technology, computer technology, communication and information technology and so on, people seek the ATCS which are suitable for the traffic characteristics of their own country. Based on the above-mentioned times background, this paper applies advanced genetic neural network to the traffic control for single intersection and urban traffic control evaluation, hoping that advanced technique can solve the existing problems in traffic control systems.The use of intelligent control technology in traffic control can be divided into two aspects, one for optimization method, the other is evaluation technology. In this paper, according to both the traffic characteristics and the present situation of traffic management, via drawing lessons from the research achievements both in china and on abroad with respect to the traffic control for single intersection and urban traffic control evaluation, confirms the analysis of genetic neutral network algorithm used in the paper. On the basis of optimal characteristics of the traffic control for single intersection, from the public transport and the common vehicles two aspects established a traffic control model based on genetic neutral network. Utilize VC programming language and vissim simulation platform to confirm the effectiveness of the model. At the same time, using the MATCS as a background, that build a traffic control model which based on genetic neutral network, and validate the efficiency of the model through the methods of programming and simulating.Chapter 1 ExordiumFirstly, introduces the research background that urban traffic problems have been the bottleneck question restricting the urban development, depending upon pure road building cannot resolve the traffic questions effectively and completely. Secondly, the paper carries on the analysis of the application condition both in china and on abroad of intelligent control technology used in traffic control systems from two aspects. On the basis of the research background, this chapter induces the significance and the goal of the paper. Finally, according to traffic control characteristics, proposes the heredity neural network concrete algorithm and the flow which will be used in this paper.Chapter 2 A study of intelligent control technology based on genetic neural network In the first place, introduces the origin, principle, optimization process as well as the advantages and disadvantages of both genetic neural network algorithm and BP Neural Network respectively. Subsequently, this chapter combines both algorithms due to their respective disadvantages and summarizes the already existing main ideas of genetic neutral network. Finally, proposes the specific algorithm and procedure which are established on the basis of traffic control characteristics.Chapter 3 Study of Single Intersection Control Based On Genetic Neural Network This chapter starts with selection of evaluation index of single intersection's control effect, it builds a single intersection control structure based on genetic neural network with a thought of transit priority, this structure evaluates intersections from the different operation condition of general vehicle and bus, if the evaluation result is negative, then adjusts the genetic neural network to get an optimal real-time control and reduce vehicle delay of the whole intersection gradually. This chapter establishes a synthesis to evaluate the functions using efficiency coefficient which can quickly evaluates the intersection's situation. Finally, the simulation experiments confirm the efficient of the proposed models.Chapter 4 Study on The Evaluation Methods of Area Traffic Control Based on Genetic Neural NetworkTo start this chapter, we analyses the MATCS system. According to the evaluation principles of the effect of the regional control, the regional average vehicle delay, the average number of stops of the regional vehicles, the regional average length of queue, and the regional average traffic capacity are used as the evaluation index. Coefficient of variation method is used to determine the objective weight of the intersections. We established a comprehensive evaluation function using the efficacy coefficient method, and establish a evaluation model for regional control on genetic neural network. By simulation, it validated that using the timing programs of MATCS system with a evaluation of regional control would be much better than that without a evaluation of regional control on the average vehicle delay.Chapter 5 Results and ProspectsThis chapter has carried on the simple summary of the research content and puts forward the proposal to the next step research work. The obtained achievements lie the following aspects:Summarizes the classifications of present heredity neural network technology, carries on the explanation to the characteristics of heredity neural network under per classification from some different respects, propose the heredity neural network technology which takes consideration of the characteristics of traffic control.Summarizes the principles used in goal selection in traffic control systems: Scientific, typical, integrity, feasibility, commensurability and so on.This paper introduces the control structure proposed in the article《AN APPROACH OF INTERSECTION TRAFFIC SIGNAL CONTROL》and establishes a genetic neural network optimization structure used in single intersection base on multi-objective evaluation criterion.Carries on the single intersection control evaluation from the two different aspects of the public transport and the common vehicles then introduces the concept of"transit priority".By means of efficiency coefficient method establishes comprehensive evaluation function applied in single intersection and urban traffic control.Adopted with variation coefficient method, per single intersection weights are determined in the urban traffic control systems.Utilize VC programming language and vissim simulation platform to confirm the effectiveness with respect to the use of genetic neural network in the traffic control system.