A PLS-SVM model for reservoir identification in natural gas ezploration
Reservoir identification plays an important role in natural gas exploration, especially difficult to accurately distinguish gas, water and dry strata in tight reservoirs, the accuracy of conventional identification methods are low. So this paper proposes a PLS-SVM model to solve the problem. SVM is very good in global optimization and generalization and is suitable for classifying different parents,. While, PLS is integrated with SVM to treat the characters of samples, in order to overcome multilinear correlation among variables and reduced the dimension of input variables at the same time. Applying the PLS-SVM model to M51 reservoir in Central Gas-field, 92 samples were selected, in which 78 samples was used to establish the model and the others to test. The result indicated that the accuracy of the PLS-SVM model reached 92.86%, compared with 85.71%, 85.71% and 78.57% of PLS-BP, SVM and BP net respectively. Therefore, the model is potential for identifying tight reservoirs and can provide useful reference for the similar research in other regions.
partial least square support vector machine PLS-SVM model reservoir identification natural gas ezploration
Kuang Jianchao Luo Xin Wang Zhong Zeng Jianyi Zhao Lu
Chengdu University of Technology, Chengdu 610059, China
国际会议
The 6th International Conference on Partial Least Squares and Related Methods(第六届偏最小二乘及相关方法国际会议)
北京
英文
84-92
2009-09-04（万方平台首次上网日期，不代表论文的发表时间）