基于形态-变长夹角链码的测井曲线识别  

Recognition of Well-logging Curve Based on Morphological-included Angle Chain of Changeable Length

作　　者：尚福华[1] 赵擎华 

机构地区：[1]东北石油大学计算机信息技术学院,大庆163318

出　　处：《计算机与数字工程》2014年第9期1587-1590,1682,共5页Computer & Digital Engineering

基　　金：国家自然科学基金资助项目(编号:61170132);国家重大专项(编号:2011ZX05020-007);黑龙江省教育厅科学技术研究资助项目(编号:12521055)资助

摘　　要：在同位素示踪注水剖面测井中,磁性定位测井曲线的形态能直观反映井下工具的类型,针对曲线整体形态相似而局部曲率、幅度、波峰个数不同的识别问题,提出了一种以曲线形态语义与变长夹角链码结合的曲线编码和描述方法。用折线重构测井曲线,以曲线的形态类型,折线的长度和相邻折线的夹角作为特征描述,采用基因遗传算法求得最优权值,对折线长度和夹角特征分别进行加权,最后用特征加权K近邻算法实现对井下管柱工具的分类。实验表明该方法能有效的用于井下工具的识别。In the isotope tracer injection profile logging, the magnetic positioning logging curves can directly reflect the type of downhole tools. To solve this identification problem that the overall shapes of the curves are similar, but other features of the curve, e. g. , the local curvature, amplitude, number of peaks are not the same, a novel approach, curve morphological combined with included angle chain of changeable length, is presented for curve encoding and representation. The curve will be represented by straight lines, and described by features about length of the polyline and the angle between adjacent polyline. The genetic algorithm is used to get the optimal weight, then polyline features and angle features are weighted based on that data. Finally, the feature-weighted K-nearest neighbors are used to classify the downhole tools. Experimental results show that this method can effectively be used to identify downhole tools.

关 键 词：变长夹角链码 形态描述 形态识别 K近邻算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]