基于斜率偏离的时间序列相似性搜索方法研究  

Study of distance-measuring method and application for time series similarity search based on slope departure

作　　者：梁建海[1] 张建业[2] 杨峰[1] 潘泉[1] 

机构地区：[1]西北工业大学自动化学院,西安710072 [2]空军工程大学,西安710051

出　　处：《计算机应用研究》2010年第1期54-55,共2页Application Research of Computers

基　　金：国家自然科学基金资助项目(60634030;60702066);国防预研基金资助项目

摘　　要：针对数据挖掘领域中时间序列的相似性度量问题,提出了基于斜率偏离度量的浮动索引相似性搜索算法。在斜率偏离度量的基础上建立分箱标志,通过浮动索引方法有效实现高维多元时间序列的准确索引,用实际的飞行数据进行仿真验证获得好的效果,证明这种方法的有效性和可靠性。Aiming at the measuring method of similarity search in time series, this paper proposed a similarity measuring method based on slope departure. To dispose the time series data with high dimension, used a similarity measuring method to caculate the slope departure based on curve-in bin solving the localization of slope distance measurement in SFFS. In fact, the physical concept of this method proved more clearly and nicely with real fight data. The maturity, validity and the continuity of slope departure was proved.

关 键 词：时间序列 斜率偏离 相似性搜索 浮动索引 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]