| 基于Agent控制器的单路口信号灯学习控制方法研究 |
| |
| 引用本文： | 高丽颖,陈阳舟,李振龙. 基于Agent控制器的单路口信号灯学习控制方法研究[J]. 交通与计算机, 2007, 25(5): 59-62 |
| |
| 作者姓名： | 高丽颖 陈阳舟 李振龙 |
| |
| 作者单位： | 北京工业大学,北京,100022 |
| |
| 基金项目： | 北京市自然科学基金 , 北京市教委科技创新平台建设项目 |
| |
| 摘 要： | 将强化学习中的Q学习算法与Agent技术相结合,构成Agent控制器对单路口的信号灯进行控制.介绍了将经验知识与Q学习算法相结合实现的Agent学习机制,提出了一种适合交通环境的强化函数,以解决单路口的动态实时控制.经仿真软件Paramics对实际路口进行仿真实验,验证了该方法具有较好的控制效果.
|
| 关 键 词： | Q学习 Agent 强化函数 Paramics仿真 Agent 控制器 单路口 信号灯 学习机制 控制方法 研究 Traffic Lights Intersection Single Method 控制效果 验证 仿真实验 Paramics 仿真软件 实时控制 动态 强化函数 交通环境 |
| 修稿时间： | 2007-07-23 |
Agent-based Learning Control Method for Single Intersection Traffic Lights |
| |
| GAO Liying,CHEN Yangzhou,LI Zhenlong. Agent-based Learning Control Method for Single Intersection Traffic Lights[J]. Computer and Communications, 2007, 25(5): 59-62 |
| |
| Authors: | GAO Liying CHEN Yangzhou LI Zhenlong |
| |
| Affiliation: | Beijing University of Technology, Beijing 100022, China |
| |
| Abstract: | In this paper, Agent technology was combined with Q-learning to make an Agent controller, which controlled the traffic lights in single intersection. The Agent learning mechanism was realized by combining experience knowledge with Q-learning. A reward function for real traffic condition was proposed to control traffic dynamically and on real-time. The performance of this method is evaluated by the traffic simulation software Paramics, and the experimental results indicate the effectiveness of the method. |
| |
| Keywords: | Q-learning Agent reward function ~ Paramics simulation |
| 本文献已被 维普 万方数据 等数据库收录！ |
