基于小波变换的地震属性多源信息融合预测技术研究

RESEARCH OF PREDICTION TECHNOLOGY IN SEISMIC ATTRIBUTE MULTI-SOURCE INFORMATION FUSION BASED ON WAVELET TRANSFORM

油气预测一直以来都是地震油气勘探的重要类容,但由于地下地质构造的复杂性,往往使得这项工作困难重重。实验提出了一种基于小波变换的地震属性多源信息融合方法,包括了属性数据的标准化处理、小波分解、数据融合与小波重构,其能够将多种属性的优势特征信息有机的结合起来,从而为油气预测提供可靠与准确的依据。为了验证本方法的有效性,将其应用于准噶尔盆地X区块的油气有利储集预测。首先对多种属性进行标准化处理,在此基础上进行小波分解,获得各属性的多种频带,进而对高低频带采用不同的方法融合,最后进行小波重构获得融合数据属性。实际资料处理效果显示,相对变换优于传统线性函数标准化处理,而基于小波变换融合方法优于传统的PCA变换融合,不仅能够消除干扰与冗余信息,而且增强了各属性的优势特征与细节信息,可以较好的改善多解性问题,提高储层预测和油气检测的成功率。

Based on wavelet transform, this paper proposes a kind of multi-source information fusion method on seismic attributes, which includes standardization of attribute data processing, wavelet decomposition,data fusion and wavelet reconstruction. It can coordinate the advantage of multiple attribute characteristic information, and provide reliable and accurate basis for reservoir prediction. This method is applied to the prediction of X block reservoir in Junggar Basin in order to verify its validity. First, wavelet decomposition is carried out after the standardization of all attribute and obtains a variety of frequency bands of each attribute,in which the high and low frequency bands are processed with different fusion methods. Finally, wavelet fusion data attributes are reconstructed. It shows that the relative transform data standardization is superior to the traditional linear function. Moreover, the fusion method based on wavelet transform overmatches the traditional PCA transform,since it can not only eliminate the interference and redundant information, but also enhance the advantage characteristics and details of each attribute. This method im- proves the multiple solution problems and promotes the success rate of reservoir prediction and hydrocarbon detection.