摘要
牛文杰 ,杨钦 .超大数据体泛克里金插值的研究 .石油地球物理勘探 ,2 0 0 1,36 (6 ) :6 72～ 6 79 超大数据体泛克里金插值时 ,漂移的形式、残余 (residual)变异函数参数的确定特别费时和困难。本文提出了在三角网格剖分下提取用于估计残余变异函数参数的特征子集 ;在三角网格剖分过程中一次确定三角形与所有数据点的包含关系 ,用于快速检索待插点邻域内的数据点 ;用多元逐步回归法确定漂移多项式次数 ;将矩法与近似最大似然法相结合估计残余变异函数参数 ,较好地解决了这一问题 ,最后给出了应用实例。
Definition of shift form and parameters of residual variation function is very time-comsuming and difficult in universal Kriging interpolation of super-giant data volume.The paper presented a method that picking characteristic subset used for evaluating the parameters of residual variation function under triangulation;the inclusion relation between trigon and all data points can be defined only once in a process of triangulation,which can use for fast retrieval of data points in an adjacent domain of non-interpolated points;times of shift polynomial can be defined by multi-element progressive regression,combining moment method with maximum likelihood of approximation to evaluate parameters of residual variation function,all these together can better solve the problem.Finally, the application cases are given in the paper.
出处
《石油地球物理勘探》
EI
CSCD
北大核心
2001年第6期672-679,共8页
Oil Geophysical Prospecting
关键词
超大数据体
矩法
最大似然法
泛克里金插值
三角网格剖分
特征子集
多元逐步回归
残余变异函数
漂移次数
super-giant data volume,moment method,maximum likelihood,universal Kriging interpolation,triangulation,characteristic subset,multi-element progressive regression,residual,variation function,times of shift
