基于三角形中线的数据序列线性拟合算法  

Data Series Linear Fitting Algorithm Based on Triangular Midline

作　　者：杜奕[1] 卢德唐[1] 李道伦[1] 张挺[1] 

机构地区：[1]中国科学技术大学工程科学软件研究所

出　　处：《计算机工程》2008年第13期21-23,共3页Computer Engineering

基　　金：国家“973”计划基金资助重点项目(2006CB705800);2007年上海高校选拔培养优秀青年教师科研专项基金资助项目(egd-07010)

摘　　要：从石油测井数据解释的实际应用需求出发,提出一种新的基于三角形中线的数据序列分段算法。在扫描数据的过程中依次计算3个连续数据形成的三角形中线长度,根据自定义的中线长度阈值选择反映序列趋势变化的关键转折点,实现数据序列的线性拟合。实验结果表明该算法具有良好的拟合质量和较高的效率。Based on analyzing the relation of data points in time series, a novel series segmenting algorithm based on triangular midline is presented. During scanning these temporal data, this approach chooses three continuous data points in turn and calculates the triangle＇s midline. According to these values and the user-defined distance threshold, this method records important turning points reflecting the sequence＇s feature. Using these key points, the original time series is segmented and fitted linearly. Experimental results show that the new method is effective.

关 键 词：数据序列 线性拟合 转折点 三角形中线 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]