|
|||||
|
|
| 基于最近邻法的短时交通流预测 | |
| 引用本文： | 周小鹏,冯奇,孙立军. 基于最近邻法的短时交通流预测[J]. 同济大学学报(自然科学版), 2006, 34(11): 1494-1498 |
| 作者姓名： | 周小鹏 冯奇 孙立军 |
| 作者单位： | 1. 同济大学,道路与交通工程教育部重点实验室,上海,200092;上海市公安局,交通警察总队,上海,200070 2. 上海科达市政交通设计院,上海,200030 3. 同济大学,道路与交通工程教育部重点实验室,上海,200092 |
| 摘 要： | 针对交通流量变化存在周期性和随机性的特点，提出一种基于最近邻法的预测方法．着重介绍了状态向量构造、近邻范围确定和权重计算方法三方面的研究．根据流量与速度、占有率的关系，认为状态向量中不必考虑速度和占有率这两个交通参数；与传统最近邻法不同，近邻的个数不设为常量，而取决于所能搜索到的记录数；通常根据距离远近赋予权重的规则不可靠，而采用了等权重法．通过实际数据检验，预测误差低于7％． |
| 关 键 词： | 短时交通流预测 最近邻 状态向量 权重 |
| 文章编号： | 0253-374X（2006）11-1494-05 |
| 收稿时间： | 2005-12-28 |
| 修稿时间： | 2005-12-28 |
Short-Term Traffic Flow Forecasting Based on Nearest Neighbor Algorithm |
|
| ZHOU Xiaopeng,FENG Qi,SUN Lijun. Short-Term Traffic Flow Forecasting Based on Nearest Neighbor Algorithm[J]. Journal of Tongji University(Natural Science), 2006, 34(11): 1494-1498 | |
| Authors: | ZHOU Xiaopeng FENG Qi SUN Lijun |
| Affiliation: | 1. Key Laboratory of Road and Traffic Engineering of the Ministry of Education, Tongii University, Shanghai 200092, China;2. Traffic Police Department of the Shanghai Public Security Bureau, Sbanghai 200070,China; 3. Shanghai Keda Municipal and Traffic Design Institute, Shanghai 200030, China |
| Abstract: | Since traffic volume has the property of periodicity and randomicity, a new forecasting method based on nearest neighbor algorithm is proposed. State vector definition, neighbor selection and weights calculation are emphases of this paper. According to the relationship among traffic volume, speed and occupy, speed and occupation are considered not necessary to the state vector. Different from traditional nearest neighbor algorithm, the number of neighbors is not constant, but dependent on the recorders that can be found in history database. It is found that the principle of weighing the selected neighbor by inverse d distance is not reasonable, and equal weight approach is adopted. The mean absolute percentage error is lower than 7 %. |
| Keywords: | short-term traffic flow forecasting nearest neighbor state vector weighing |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ | |