Huang Yaping, Yang Haijun, Qi Xuemei, Malekian Reza, Pfeiffer Olivia, Li Zhixiong
School of Resources and Geosciences, China University of Mining and Technology, Xuzhou, China.
Key Laboratory of Coal Methane and Fire Control, Ministry of Education, China University of Mining and Technology, Xuzhou, China.
PLoS One. 2018 Feb 2;13(2):e0192407. doi: 10.1371/journal.pone.0192407. eCollection 2018.
The selection of seismic attributes is a key process in reservoir prediction because the prediction accuracy relies on the reliability and credibility of the seismic attributes. However, effective selection method for useful seismic attributes is still a challenge. This paper presents a novel selection method of seismic attributes for reservoir prediction based on the gray relational degree (GRD) and support vector machine (SVM). The proposed method has a two-hierarchical structure. In the first hierarchy, the primary selection of seismic attributes is achieved by calculating the GRD between seismic attributes and reservoir parameters, and the GRD between the seismic attributes. The principle of the primary selection is that these seismic attributes with higher GRD to the reservoir parameters will have smaller GRD between themselves as compared to those with lower GRD to the reservoir parameters. Then the SVM is employed in the second hierarchy to perform an interactive error verification using training samples for the purpose of determining the final seismic attributes. A real-world case study was conducted to evaluate the proposed GRD-SVM method. Reliable seismic attributes were selected to predict the coalbed methane (CBM) content in southern Qinshui basin, China. In the analysis, the instantaneous amplitude, instantaneous bandwidth, instantaneous frequency, and minimum negative curvature were selected, and the predicted CBM content was fundamentally consistent with the measured CBM content. This real-world case study demonstrates that the proposed method is able to effectively select seismic attributes, and improve the prediction accuracy. Thus, the proposed GRD-SVM method can be used for the selection of seismic attributes in practice.
地震属性的选择是储层预测中的关键环节，因为预测精度依赖于地震属性的可靠性和可信度。然而，有效选择有用地震属性的方法仍然是一个挑战。本文提出了一种基于灰色关联度（GRD）和支持向量机（SVM）的储层预测地震属性选择新方法。该方法具有两级结构。在第一级中，通过计算地震属性与储层参数之间的灰色关联度以及地震属性之间的灰色关联度来实现地震属性的初选。初选原则是，与储层参数灰色关联度较高的地震属性之间的灰色关联度，相较于与储层参数灰色关联度较低的地震属性之间的灰色关联度更小。然后在第二级中使用支持向量机，利用训练样本进行交互式误差验证，以确定最终的地震属性。进行了一个实际案例研究来评估所提出的GRD-SVM方法。选择了可靠的地震属性来预测中国沁水盆地南部的煤层气（CBM）含量。分析中，选取了瞬时振幅、瞬时带宽、瞬时频率和最小负曲率，预测的煤层气含量与实测煤层气含量基本一致。这个实际案例研究表明，所提出的方法能够有效地选择地震属性，并提高预测精度。因此，所提出的GRD-SVM方法可用于实际中的地震属性选择。
