最大隶属优势准则在测井岩性识别中的应用

The Application of the Criteria of the Biggest Membership Advantage to the Recognition of Weil Logging Lithology

在测井岩性识别过程中,被测岩性是一个模糊概念,因此,可用模糊式识别法进行岩性识别。设测井地区地层岩性有m个状态,用模糊子集(i=1,……,m)表示。又设与地层岩性有关的观测量有n个,每次观测结果为一个n维向量X=(X_1,X_2,……X_n)。于用X可以描述钻井所穿过的地层的岩性状态。 U_(X)表示X从属于岩性Ai的程度。一般模糊模式别是求出诸U_(Xi),i-1,……,m)。若U_(X)=max{U_(X),1≤mi≤},则判定X属于Ai,从而可知与X对应的地层的岩性。此即最大隶属原则在测井岩性识别中的应用。其关键是求出U_(X)。然而,由于模糊集运算贫乏,丢失信息多,使求U_(X)的方法均不理想。本文则避开直接求U_(X),而引入隶属优势概念和最大隶属优势准则,对测井岩性进行模糊式识别。本文提出了一种测井岩性的模糊模式识别新方法。此法的关键是如何求U_(X_i),(i=1,……m;j=1,……n)。文中还介绍了两种求U_(X_i)的方法。

In the process of the recogniton of well logging tithology,the well logging lithology is rather a blurring concept; therefore it is possille to employ the method of fuzzy pattern recognition to carry out the well logging lithology recognition. Suppose that the states of the lithology of the strata in the well logging area is m, which can be expressed by fuzzy sets (i=1, 2…m ). Again we assume that the observation value in relation to the lithology of the strata is n, and therefore the result of each observation is a n dime- nsionnality vector, indicated by x=x_1, x_2…x_n Then X can be used to descibe the states of the lithology of the strata which have been penetrated by drilling. U_(X_i)indicates the degree to which X is subordinated to the lithology Ai, while the aim of fuzzy pattern recognition is to work out each U(X_i), (i=1…m)。IfU_(X_i)=max{Ai(X), 1≤i≤m}, we know that X belongs to and can further predict the lithology of th strata in correspondance to X. The crucial point is to work out U_ (X_i). However, the approaches applied to get U_(X_i)have much to be desired because funny sets have limited ways of operation and inevitably lose a lot of indispensable information. This article deliberately steers clear of the attempt to get U_(X_i), but introduces the membership advantage and the criteria of the biggest membership advantage so as to apply fuzzy pattern recognition to well logging lithology. Upon the basis of the principle mentioned above, this article puts forward a new method of fuzzy pattern recognition for weel logging lithplogy. The crucial point of the method lies in how to work out U_ (X_i), (i=1,….m, j=1…n). Movever, the article also introduces tow practical ways of getting U(X_i).