提高采收率潜力分析的基础是进行提高采收率方法的潜力预测 .建立提高采收率潜力预测模型从统计学习的角度来看 ,实质是属于函数逼近问题 .本文首次将统计学习理论及支持向量机方法引入提高采收率方法的潜力预测中 .根据 Vapnik结构风险最小化原则 ,应尽量提高学习机的泛化能力 ,即由有效的训练集样本得到的小的误差能够保证对独立的测试集仍保持小的误差 .在本文所用较少样本条件下 ,支持向量机方法能够兼顾模型的通用性和推广性 ,具有较好的应用前景 .研究中采用的是综合正交设计法、油藏数值模拟和经济评价等方法生成的理论样本集 Based on the statistical learning, the essence is the function approximation problem.This paper first introduces the theory and support of statistical learning According to Vapnik’s principle of minimizing structural risk, the generalization ability of learning machine should be improved as much as possible, that is, the small error obtained from effective training set samples can guarantee the independence of testing Set remains a small error.In this paper, the use of fewer sample conditions, support vector machine method can take into account the versatility and generalization of the model, has a good application prospects.In this study, the integrated orthogonal design method, the reservoir Numerical simulation and economic evaluation methods to generate the theoretical sample set