摘要:
基于模糊式识别,本文提出聚类分析与贴近度识别相结合应用于海相碳酸盐岩岩相的分类和识别,聚类用F—PFS法,识别用本文建议的一种带参数的名为加权逆距离贴近度的新公式:S（Xi,Yj）=sum from k=sA-1∑ Wk(1-Dijk)p。其中权重Wk使用一种按单因素判对率确定的新技术。该权重并可用于模糊综合评判。模糊特征向量采用扩展的正则化式计算。上述方法在湖北三叠系海相碳酸盐岩地层得到了应用,并将其划分为六个相.选用12—21个特征因素（颜色、矿物成分、古生物、颗粒组分、填隙物、沉积构造等）,并定量化赋值。聚类分析可分出四个相,贴近度识别的判对率达81.8%,按划分为六个相的贴近度识别,确切判对率达79.2%。
Abstract:
On the basis of fuzzy pattern recognition, a practical method of recognition of marine carbonatite fades is presented in this paper. The standard pattern can be constructed in the situation of cluster analysis and known samples. The cluster centre is obtained by the F-PFS method or the characteristic vector. Besides of the using of cluster analysis, the recognition of unknown sample mainly uses the calculation of similarity degree and the recogition is according to the nearest neighbour criterion. Then, a new similarity degree of complement distance with the different weights is presented in the paper. That is S(Xi, Yj)=sA-1 Σ Wk(1-Dijk)p (2) k - I Where the weight Wk is determained by the technique which is presented by the authors and is according to the rate of right judgement of the single factor. Four facies can be identified by using fuzzy cluster analysis, other samples can be recognized by using the fuzzy recognition method. The percent of right judgment is 81.8%. Six standard patterns were built based on the known facies samples of Shouzi Section. Both the samples of Jingshan Section and the rest samples of Shouzi Section were recognized with this model, the percent of right judgment is 79.2%. The results of fuzzy recognition show that fuzzy cluster analysis and similarity degree are a kind of new effective method for the classification of carbonatite facies. Furthermore, cluster centres based on the fuzzy cluster or known facies samples could become the basis of later fuzzy recognition marine carbonatite facies.
点击查看大图
下载:
