测井时间序列的计算机沉积微相识别，在特征提取上有一定的难度。证明测井时间序列是混沌的，是提取其混沌特征的前提条件。混沌时间序列的判定目前已有不少方法，本文结合测井曲线的特点，利用相空间重构技术和G—P算法，用关联维数方法证明了测井时间序列确实存在混沌。实验数据来自某油田的实际测井数据。实验结果表明，油层组的关联维数与油层的构造特点有关，并发现了该区域油井的油层组关联维数一般大于全井这一有趣的现象。

Oil bearing layers and their sedimentary microfacies are generally analyzed by collecting oil well logs in the oilfield development. The well log time series is the correct reflection of geological characteristics of the oil bearing layers with quite a high probability. Unfortunately, feature extraction is of certain difficulty to the computer sedimentary microfacies recognition with well log time series. Oil bearing layers are the resultants of sedimentary layer sequence. The formation of oil bearing layers is very complicated and variable. Extracting the chaotic features is promising. But proving that well log time series are chaotic is prerequisite to extract their chaotic features. There are several methods developed for the chaos identification at present. In this paper, the well log time series is proved to be chaotic indeed by phase space reconstruction technology and G P algorithm, which is a practical algorithm for computing the correlation dimension of a time series. The experiment results indicate that the correlation dimension of an oil layer group is related to its structure. Another interesting phenomenon is that the correlation dimension of an oil layer group is generally larger than that of the whole well.