基于支持向量机的川中杂卤石分类识别研究

发布时间：2018-07-03 04:02

本文选题：杂卤石 + 支持向量机　；参考：《地球科学进展》2016年10期

【摘要】：杂卤石是四川盆地主要的固态钾矿物,川中地区大多数杂卤石层不纯,通常伴随石膏层、硬石膏层、盐岩层发育,甚至同层沉积,常规测井解释方法只能粗略地识别杂卤石层。以支持向量机理论和测井解释为基础,测井数据作为输入,构建预测模型,对川中地区下中三叠统杂卤石样本做精细识别,将识别结果与录井资料验证对比,正确率达到90%以上。再以预测模型为基础,结合含杂卤石岩性在测井曲线上的响应情况,构建杂卤石层分类识别模型,识别杂卤石层、石膏质杂卤石层和杂卤石膏岩层,识别正确率达到91.78%,与常规测井解释方法相比具有明显优势。结果表明,将支持向量机运用到找钾矿中具有广阔的前景。
[Abstract]:Complex halite is the main solid potassium mineral in Sichuan basin. Most of the mixed halite layers in central Sichuan are impure, usually accompanied by gypsum layer, anhydrite layer, salt rock formation, and even the same layer deposition. The conventional logging interpretation method can only roughly identify the mixed halite beds. Based on the support vector machine theory and log interpretation, the prediction model is constructed by using log data as input, and fine identification of the lower middle Triassic mixed halogen samples in central Sichuan is made, and the recognition results are verified and compared with the logging data. The accuracy is over 90%. Based on the prediction model, combined with the response of the complex halogenite on the logging curve, the classification and recognition model of the complex halogenite layer is constructed to identify the mixed halite layer, the gypsum mixed halite layer and the complex halogen gypsum stratum. The recognition accuracy is 91.78, which is superior to the conventional logging interpretation method. The results show that the application of support vector machine in potassium prospecting has a broad prospect.
【作者单位】：西南石油大学地球科学与技术学院;川庆钻探工程有限公司地球物理勘探公司;中国石油新疆油田分公司勘探开发研究院;
【基金】：国家自然科学基金项目“四川盆地油钾兼探的地球物理评价方法研究”(编号:41372103) 国家重点基础研究发展计划项目“四川三叠纪古特提斯海盆钾分布、评价研究”(编号:2011CB403002)资助~~
【分类号】：P619.211

【相似文献】