基于数理统计理论的两种分层方法的比较

Comparison Between Mahalanobis Distance Judgement and Fisher Criterion Based on Math Statistic Theory.

王瑞平，白康生，赵文杰，等．基于数理统计理论的两种分层方法的比较．测井技术，１９９８，２２（５）：３３２～３３６随着地下资源的枯竭和测井技术的发展，人们期待着更精确更可靠地寻找到储集层的方法。介绍并利用费歇尔（Ｆｉｓｈｅｒ）判据分类方法识别划分目的层，并把这种方法与马哈诺比斯（Ｍａｈａｌａｎｏｂｉｓ）距离分层方法加以比较。两种方法的联合应用使分层具有“可比性”。两种方法依据多元统计理论，对“标准”传统分层资料进行随机抽样，由此可进行误差分析，判断资料的可靠性和方法的实用性。对胜利油田某口井作了实际处理，效果较好。

With exhaustion of the underground resources and development of logging technology, it is expected to find mineral formation (or bed of interest) precisely and reliably. Fisher criterion is introduced and used to identify the bed of interest, it is compared with Mahalanobis distance zonation method. The application of the two methods makes zonation verifiable. Random sampling of the “standard” traditional zonation data is made on the basis of multivariate statistical analysis theory. Thus, error analysis can be made to determine whether the data is reliable and whether the method is practical. Log data from a well in Shengli Oilfield have been processed with the two methods, the results of which are good.