摘要
油气勘探风险的定量评价一直是国内外研究的难点。作者在分析传统风险评价方法的优缺点、剖析偏最小二乘法和最大熵法优势的基础上,首次提出了偏最小二乘-最大熵(PLSME)风险分析模型。偏最小二乘法较好地实现了多元线性回归、主成分分析和典型相关分析的有效综合,通过自变量的PLS回归线性处理,不仅能消除粗差解决变量之间的相关性问题,而且能辨识每一个自变量对因变量的控制程度;最大熵法通过对偏最小二乘得出的风险因子与总经济效益净现值关系式的检验,利用最大值、最小值和最可能值的训练,能了解指标最终服从的概率分布,客观得出风险的大小。两者结合起来构建的PLSME模型,能使风险评价结果更加准确、合理和客观。通过对四川德阳新场气田的实例应用,表明偏最小二乘-最大熵评价方法科学可行,对同类研究具有借鉴作用。
A quantitative evaluation of the oilgas exploration foreign researches. maximum entropy disadvantage of th regression method This （ME） e tradi and th article puts fo risk analysis tional methods rward the partial 1 model for the firs and axmum entropy risk has been a difficulty in domestic and east squares （PLS） regression and the t time on the basis of analyzing the anatomizing the advantage of the partial least squares method. The partial least squares regression preferably realizes the integration of multiple linearity regression, principal component analysis and enterprise information maturity. By means of managing the independent variable by the PLS linearity regression, it can not only solve the problem between the variables and eliminate the outliers, but also distinguish every restraining degree from independent variable to dependent variable. Nevertheless, it will know the ultimate obedience of index probability distribution and the sizes of risk objectively by using the maximum entropy to checkout the relations of risk factors and economic NPV of total returns which get from the partial least squares regression and the maximum entropy and training making use of maximum value, minimum value and the most possible value. The structure of this PLSME model combined with these two methods could adapt to the risk characteristic of oil-gas exploration and make the result of the risk evaluation more accurate, rational and objective. By using the example of the Xinchang gas field in Sichuan, this article indicates that the partial least squares regression and the maximum entropy is feasible and useful for reference to the congener research.
出处
《成都理工大学学报（自然科学版）》
CAS
CSCD
北大核心
2008年第1期52-56,共5页
Journal of Chengdu University of Technology: Science & Technology Edition
关键词
油气勘探
风险评价
偏最小二乘
最大熵法
PLSME模型
oil-gas exploration
risk analysis
partial least squares regression（PLS）
maximum entropy（ME）
PLSME model
