摘要
依据时间序列的形态特征,为基于欧氏距离的相似性度量加入奖惩因子,使其能反映序列形态的相似性。同时根据相关的背景知识给时间序列不同的维设定不同的权重,并给出一种自动求权重集合的算法。该算法成功应用于福泉高速行车数据以及仿真数据的相似性度量。
In this paper we added the rewards and punishment factor into the Euclidean similarity metrics according to the morphological characteristics of time series to enable it being able to represent the similarity of the series morphology. Besides, we set different weight to dif- ferent dimensions of time series according to corresponding background knowledge, and proposed an algorithm for automatic weight set search. The algorithm has been applied to the similarity metrics of Vehicle Traffic Data of the Fuquan Expressway and its simulation successfully.
出处
《计算机应用与软件》
CSCD
2010年第9期116-118,共3页
Computer Applications and Software
基金
福建省青年创新基金(2006F3075)
关键词
时间序列
权重
形态
相似性
Time-series Weight Morphology Similarity
