基于分段时间序列的试井数据相似性查询算法  

Querying Similar Well-Logging Data Based on Segmented Time Series

作　　者：杜奕[1] 李纪锋[2] 

机构地区：[1]上海第二工业大学计算机与信息学院,上海201209 [2]中国科学技术大学工程科学软件研究所,合肥230027

出　　处：《上海第二工业大学学报》2008年第2期97-102,共6页Journal of Shanghai Polytechnic University

基　　金：2007年上海高校选拔培养优秀青年教师科研专项基金(No.egd-07010)资助项目

摘　　要：从石油试井数据序列查询的实际需求出发,给出一种新的时间序列相似性查询算法。该算法首先通过中线距离阈值和极值点两个约束条件分段线性拟合时间序列,利用分段动态弯曲距离度量获得相似的分段子序列,逐点检索该子序列实现序列的精确查询。实验结果表明,该算法具有良好的相似性查询质量和效率。A new algorithm for querying similar time series was proposed to satisfy the requirement of searching similar well logging data in oil field. The approach first segments time series using two parameters of midline distance and turning points. After finding similar segmented subsequence by Segmented Dynamic Time Warping Distance, this method then search this subsequence point by point, by which the subsequence is acquired accurately. The experimental results show that this algorithm is efficient.

关 键 词：时间序列 分段拟合 分段动态弯曲距离 相似性查询 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]