利用测井数据进行岩性识别的多元统计方法

在中国大陆科学钻探CCSD-1井110～2 000 m井段实测资料的基础上,通过取芯井岩芯和相应测井曲线对应特征的分析,挑选出对岩性识别能力强的测井参数,应用多组判别分析方法,建立了研究区各类岩性的识别模型.应用该模型可对井剖面地层进行连续自动解释,获得整口井的岩性解释剖面图,为进一步的地质研究提供基础资料.最后将得出的结果与分层资料进行误差分析,以判断资料的可靠性和方法的实用性.

Multivariate Statistical Method of Utilizing Logging Data to Lithologic Recognition

Lithology identification is very important in such geological studies as stratigraphic correlation and sedimentary facies analysis. While multivariate statistical analysis is, in particular, a very effective method to recognize the lithology of strata when there are less core data and more well log data. On the basis of the CCSD-1 well data of 110-2 000 m and the analysis of the correlative characters of drilling cores and well logs,some well log parameters useful for recognizing lithology are selected. Then the function of lithological recognition for each type is set up by using multivariate discriminative analysis method. The function can be applied for the automatic explanation of vertical stratum and the lithological explanation for the whole well. Finally,the results are compared with the data of stratum classification to judge the dependatility of the data and practicability of the methods.