摘要
油层识别是石油勘探与开发的主要任务,也是测井解释的主要工作之一.目前基于支持向量机(SVM)的油层识别方法比较盛行,如基于LS-SVM的方法,但它存在惩罚参数和所选核函数的宽度参数不易选取问题,为此,可以采用量子粒子群算法(QPSO)优化LS-SVM,提出了基于QPSO的LS-SVM油层识别方法,主要包括测井数据预处理,样本数据选取、属性约简、支持向量机建模和实际油层识别等步骤.实际资料处理表明,所提出的油层识别方法优于常用LS-SVM油层识别方法,其应用效果显著.
Oil reservoir recognition is one of the main tasks in oil exploration and development, also the one of main woks in well logging interpretation. At present, reservoir identification methods based on support vector machine(SVM) are very popular, such as the method based on least squares support vector machine(LS-SVM) but it is difficult to select the penalty parameter and the kernel width parameter in LS-SVM, so the quantum particle swarm optimization algorithm(QPSO) is adopted to optimize the LS-SVM, and the reser-voir identification method based on the QPSO and LS-SVM is presented, which mainly includes logging data processing, sample data selection, attribute reduction, SVM modeling and the actual reservoir identification. The actual data processing indicates that the presented method for reservoir identification is superior to the commonly method with LS-SVM and its application effect is remarkable.
出处
《河北工业大学学报》
CAS
北大核心
2013年第4期4-8,共5页
Journal of Hebei University of Technology
基金
国家自然科学基金(60972106
51208168)
天津市自然科学基金(11JCYBJC00900
13JCYBJC37700)
河北省自然科学基金(F2013202254
F2013202102)
河北省引进留学人员基金(C2012003038)
