摘要
在对交通流量进行短期预测时,历史数据的选取至关重要.定义了相似系数和波动系数,分别对每周交通流量相似性和不同周同一工作日的交通流量相似性进行分析,在对周期相似性判定的基础上,选取相似的交通流量数据作为训练数据.利用LSSVM模型,以1d为1步,进行连续5步的交通流量预测,实践结果表明LSSVM在交通流预测领域具有良好的适应性及应用前景.
The selection of historical data is essential to the short-term traffic flow prediction. This paper defines a similarity coefficient and a fluctuation coefficient. The traffic flow simi- larity of each week and the same working day in different weeks are analyzed respectively. On the basis of cycle similarity determinant, the similar traffic flow data are selected as the training data. During the prediction process of traffic flow using LSSVM, a day as a step, the 5-step prediction is done in succession. The results show that the model has better generalization ability. Therefore,there is a favorable application prospect.
出处
《青岛理工大学学报》
CAS
2013年第2期86-91,共6页
Journal of Qingdao University of Technology
关键词
交通流量
流量预测
相似系数
最小二乘支持向量机
traffic flow
flow prediction
similarity coefficient
least squares support vector machines（LSSVM）
