Mobile version of Hanspub

文章引用说明 更多>> (返回到该文章)

许建华, 蔡瑞. 有监督SOM神经网络在油气预测中的应用[J]. 石油物探, 1998, 37(1): 71-76

被以下文章引用:

  • 标题: SVM岩性识别方法在准噶尔盆地物性模板构建中的应用Application of SVM-Based Lithology Identification in Building Rock Physical Property Templates of Junggar Basin

    作者: 于会臻, 王树华, 郭涛

    关键字: 支持向量机, 岩性识别, 物性模板, 准噶尔盆地SVM, Lithology Identification, Physical Property Template, Junggar Basin

    期刊名称: 《Advances in Geosciences》, Vol.6 No.1, 2016-02-25

    摘要: 岩性识别是构建岩石物性模板、开展重磁电震联合反演解释工作的基础。通常采用的交汇图方法进行岩性识别,该方法根据判断交汇点平面分布差异性来给出岩石岩性识别标准,易受主观因素影响,且无法直接应用于多于三类物性数据的岩性识别,尤其在岩石物性规律复杂时识别精度更低。针对该问题,基于支持向量机理论提出了一种更稳定的重磁电震物性值岩性识别方法,以此为基础构建了准噶尔盆地的岩石物性模板,并在车排子地区的岩性判别中得到了较好的应用。 It is a fundamental work to identify lithology for building rock physical property template and carrying out gravity-magnetic-magnetotelluric-seismic joint inversion & interpretation. Usually we use the cross-plot maps to identify lithology. Its lithology identification standard is given according to the difference of plane distribution of intersection point. However, this method has low accuracy because it is vulnerable to subjective factors, and can’t be used in high-dimensional classification; especially if physical property data is complicated. To solve the problem, based on support vector regression machine theory, this paper proposes a more stable lithology identification method, and sets up multiple physical property templates of variant rocks of Junggar basin, and it has been successfully applied to lithology discrimination of Chepaizi area.