测井时间序列的替代数据混沌判定方法的研究

Study of the surrogate data method for chaos identification of well-log time series

利用测井时间序列进行计算机自动测井解释已成为正确认识油层地质规律的重要途径,但能够有效区分沉积相的曲线特征却难以提取.由于测井时间序列的非均质性,线性特征已不能满足逐渐细分的多种沉积微相的分类要求.而提取时间序列的非线性(混沌)特征时,首先要判定它是否含有混沌特性.这里将替代数据法应用于某油田的3种真实测井数据的混沌判定,并结合全井的关联维数做判据进行实验.得出的判据显示,3种测井时间序列均含有混沌.这一结果表明,具有复杂形态的测井时间序列可以用低阶非线性模型或适当的分维特征来描述.这为进一步应用混沌时间序列分析方法研究测井数据打下了理论基础.

Automatic logging data interpretation using well-log time series becomes a significant way of obtaining the correct geological laws of oil-layers. But extracting the effective pattern features for sedimentary facies classification is a problem. Linear features can't satisfy the finer classification of sedimentary microfacies. When extracting nonlinear (chaotic) features, it is necessary to consider if the original time series contains a certain amount of chaos. There are several methods for chaos identification such as Lyapunov exponents and correlation dimensions. The surrogate data method perfects the embedding space theory and becomes a more reliable one for chaos identification by combining other quantitative features. In this paper, the surrogate data method is introduced to identify chaos in three kinds of real well-log data of an oilfield in China. Combining the correlation dimension of the whole well-log, every identification evidence S is larger than 1.96. This indicates that the well-log time series of the oilfield are chaotic. So the pattern-complicated well-logs can be represented by nonlinear models of low orders or certain fractal dimensions. This provides the theoretical foundation for further study of well-log data interpretation with the chaotic time series analysis.