Font Size: a A A

Regional Traffic Signal Control Based On Multi-agent System

Posted on:2007-02-15Degree:MasterType:Thesis
Country:ChinaCandidate:Y F MaFull Text:PDF
GTID:2209360182497416Subject:Management Science and Engineering
Abstract/Summary:PDF Full Text Request
Traffic problem is an important problem all over the world. It has seriously influenced the society, how to solve it is the key problem people pay attention to, now,"intelligent transportation system", which aims to solve the traffic problem using the advanced science and technology, is becoming and distributed artificial intelligence, regarded as the development of the object-oriented, provides a new ways for the research of complicated systems and is the focus of computer science, it also gives a new hopeful direction for traffic problem. Therefore, according to this traffic situation and the newly agent technology, along with the intelligent control technology, this thesis discusses a new method to deal with traffic problem.Firstly, this paper analyses the development history of the traffic signal control and the trend of traffic control system development. With the explaining of the theory of agent, the paper gives a conclusion of the conception, feature, mould and composition of agent. Next, the paper introduces the classifications and methods of traffic signal control, and stresses the main parameters and the estimation indexes of the signal control of isolated intersection.Urban transportation is very complex. For its autonomy and cooperation, agent provides a better way to take the simulation of the traffic system than others. This paper summarizes the construction of the multi-agent-based district traffic signal control system. On the summary the paper brings up a two-level model of district signal control system. Q-learning method is an online reinforcement learning recently. This paper simply introduces the main theorem and the suitable condition, and gives the steps for online method. The paper gives a method of single intersection traffic signal control and district traffic signal control using Q-learning method. This paper describes the conditions,calculate step involved in the new method, and discuss the feasibility and the validity of the new method.It is an effective way to verify the performance of system according to the test and data sampling. The method of design agent using MFC is introduced in the paper. And the paper explains the definitions,working flows and the cooperative relations...
Keywords/Search Tags:multi-agent, traffic signal control, district, Q-learning
PDF Full Text Request
Related items