本文選題：人工智能　切入點(diǎn)：多Agent系統(tǒng)　出處：《長(zhǎng)沙理工大學(xué)》2008年碩士論文　論文類(lèi)型：學(xué)位論文

【摘要】： 隨著國(guó)民經(jīng)濟(jì)的不斷增長(zhǎng),人民生活水平日益提高,汽車(chē)保有量也隨之增長(zhǎng),隨之而來(lái)的城市交通問(wèn)題則不斷突顯出來(lái)。城市交通系統(tǒng)是一個(gè)具有隨機(jī)性、不確定性、實(shí)時(shí)變化的復(fù)雜大系統(tǒng),采用以往的交通控制方式,已經(jīng)無(wú)法解決日益嚴(yán)峻的交通問(wèn)題,采用現(xiàn)代科技手段及智能方法來(lái)解決城市交通問(wèn)題成為當(dāng)前研究的熱點(diǎn)。 隨著分布式人工智能技術(shù)的發(fā)展,Agent技術(shù)和多Agent系統(tǒng)理論成為研究分布式計(jì)算環(huán)境下軟件智能化的基礎(chǔ),由于城市交通控制固有的分布性,采用多Agent技術(shù)研究城市交通信號(hào)控制問(wèn)題具有很好的前景。 本文結(jié)合Agent技術(shù)及多Agent系統(tǒng)理論,提出了以單個(gè)路口交通信號(hào)控制Agent為基礎(chǔ)的多Agent分布式協(xié)調(diào)控制系統(tǒng)。首先,對(duì)系統(tǒng)中單路口的交通信號(hào)控制Agent進(jìn)行了結(jié)構(gòu)設(shè)計(jì),對(duì)其工作過(guò)程進(jìn)行了描述,并對(duì)其學(xué)習(xí)單元的學(xué)習(xí)模式進(jìn)行了設(shè)計(jì),在此基礎(chǔ)上,對(duì)基于多Agent的交通信號(hào)控制系統(tǒng)的結(jié)構(gòu)進(jìn)行了設(shè)計(jì),并對(duì)系統(tǒng)進(jìn)行了形式化描述。接著,對(duì)單個(gè)路口的交通狀態(tài)加以選擇,采用模糊聚類(lèi)方法對(duì)車(chē)輛到達(dá)及信號(hào)顯示狀態(tài)進(jìn)行了定量描述,建立了信號(hào)控制規(guī)則集,以總停車(chē)延誤為控制目標(biāo),采用改進(jìn)的Q-學(xué)習(xí)算法對(duì)Agent進(jìn)行訓(xùn)練,以改進(jìn)信號(hào)控制規(guī)則,通過(guò)仿真,對(duì)文中提出的單路口控制方法進(jìn)行了仿真。仿真結(jié)果表明,該方法優(yōu)于傳統(tǒng)的定時(shí)控制和感應(yīng)控制方式。最后,著重對(duì)基于多Agent的分布式信號(hào)控制系統(tǒng)中,各信號(hào)控制Agent間的信息交互,協(xié)調(diào)方式,及系統(tǒng)的學(xué)習(xí)方式進(jìn)行了描述,并對(duì)系統(tǒng)協(xié)調(diào)的實(shí)現(xiàn)進(jìn)行了仿真,仿真結(jié)果表明,文中方法能明顯的減少車(chē)輛的總延誤時(shí)間。
[Abstract]:With the continuous growth of the national economy, people's living standards are improving day by day, the number of cars is also increasing, and the following urban traffic problems are constantly highlighted. The urban transportation system is a kind of random and uncertain. The complex large-scale system with real time change has been unable to solve the increasingly serious traffic problems by using the former traffic control mode. It has become a hot spot to solve the urban traffic problems by modern scientific and technological means and intelligent methods. With the development of distributed artificial intelligence technology, agent technology and multi-#en0# system theory become the basis of studying software intelligence in distributed computing environment, because of the inherent distribution of urban traffic control. It is very promising to study the problem of urban traffic signal control by using multi-Agent technology. Based on the Agent technology and the theory of multiple Agent system, this paper presents a multi-#en3# distributed coordinated control system based on a single intersection traffic signal control Agent. Firstly, the traffic signal control Agent of a single intersection in the system is designed. The working process of the system is described, and the learning mode of the learning unit is designed. On this basis, the structure of the traffic signal control system based on multiple Agent is designed, and the system is formalized. The traffic state of a single intersection is selected, the vehicle arrival and signal display states are described quantitatively by fuzzy clustering method, and the signal control rule set is established, with the total parking delay as the control target. The improved Q- learning algorithm is used to train the Agent to improve the signal control rules. The simulation results show that the proposed single intersection control method is simulated. This method is superior to the traditional methods of timing control and induction control. Finally, the information exchange, coordination mode and learning mode of each signal control Agent in the distributed signal control system based on multiple Agent are described. The simulation results show that the method can obviously reduce the total delay time of the vehicle.
【學(xué)位授予單位】：長(zhǎng)沙理工大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2008
【分類(lèi)號(hào)】：U491.51

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 承向軍,楊肇夏;基于多智能體技術(shù)的城市交通控制系統(tǒng)的探討[J];北方交通大學(xué)學(xué)報(bào);2002年05期

2 高陽(yáng),周志華,何佳洲,陳世福;基于Markov對(duì)策的多Agent強(qiáng)化學(xué)習(xí)模型及算法研究[J];計(jì)算機(jī)研究與發(fā)展;2000年03期

3 陸小芳;鄭應(yīng)平;王令群;;交通信號(hào)控制系統(tǒng)的多agent協(xié)調(diào)研究[J];計(jì)算機(jī)工程與應(yīng)用;2006年02期

4 李建民;石純一;;DAI中多Agent協(xié)調(diào)方法及其分類(lèi)[J];計(jì)算機(jī)科學(xué);1998年02期

5 徐晉暉;石純一;;多Agent系統(tǒng)中的聯(lián)盟形成[J];計(jì)算機(jī)科學(xué);1999年04期

6 楊鯤;翟永順;劉大有;;Agent:特性與分類(lèi)[J];計(jì)算機(jī)科學(xué);1999年09期

7 胡旦華,馬永光,張宇晴;多Agent系統(tǒng)中合作策略的研究[J];計(jì)算機(jī)仿真;2004年03期

8 范波,潘泉,張洪才;一種基于分布式強(qiáng)化學(xué)習(xí)的多智能體協(xié)調(diào)方法[J];計(jì)算機(jī)仿真;2005年06期

9 俞崢,李建勇;多智能體在交通控制系統(tǒng)中的應(yīng)用[J];交通運(yùn)輸工程學(xué)報(bào);2001年01期

10 劉金琨,王樹(shù)青;多智能體控制系統(tǒng)的設(shè)計(jì)與實(shí)現(xiàn)[J];控制理論與應(yīng)用;1999年04期

，

本文編號(hào)：1637370