基于贝叶斯网络的停车行为分析.pdfVIP

第 30~接第 5 期 系就工程理论句实践 Vo1.30 , No.5 201肘手 5 月 Systems Engineering … Theory Practice May, 2010 文章编号: 100也6788(2010)05-0948阳08 中回分提号: U49 1.l 文献标志码:A 基于贝叶斯网络的停年行为分析 来芳1 ，张慧永 1 ，隽志才 9 (1.吉林大学交通学院，长春 130022; 2. 上海交通大学安泰经济与管理学院，上海 200052) 摘厦建立贝叶斯网络模型进行停车行为分析.应用 K2 算法和贝叶斯参数估计方法，进行了 贝叶斯网络的结构和参数学习，建立了停车行为分析贝叶斯网络，应用联合树推理引擎推断在出行 目的、停车费率等因素的影响r停车行为的变化.结果表明，贝叶斯网络能够直观表现决策行为与 其影响因素之间的交互作用机理，并方便进行决策行为的敏感性分析，间时模型具有较高的精确度. 可以为政府和规划部门深入了解居民的停车选择行为和停车决策机理提供依据. 关键词贝叶斯网络;停车行为; K2 算法;非集计模型 Bayesian network皿based parking behavior analysis 1 1 2 ZONG Fang , ZHANG Hui-yong , JUAN Zhi-cai (1. College of Tra侃c， Jilin University, Changchun 130022, China.; 2. College of Anta.i Economics ; Ma.n品.gement， Sha.ngha.i Jia.otong University, Sha.ngha.i 200052 , China.) Ab翻tract In this paper , a Bay，础ian network wωdeveloped to analyze the parking behavior. A Bayesian network for parking behavíor analyzing w，制 built by structure and parameter learni吨， using K2 method and Bayesi皿method respectively. The influence自 of some factor日， such as the trip purpose and the parking f例， on the parking behavior were computed using the junction tree engine. The r暗自ults indicate that the Bayesian network can expr剧目 the ωupling mechanism between the d创ision behavior 阳d the influencing factors visually, and c刷刷也lyze the sensitivity of the decision behavior conveniently. MorωIver， ， the Baye日ian network has a high accuracy degree. The re自ults of this paper can provide basis for the government to study resident日， parking behaviors and parking decision mechanism. Keyword自 Bayesian networkj parking behavior; K2