摘要
储层损害快速预测模型的建立可以及时准确地提供储层潜在损害类型及程度,为储层保护工作指明方向。然而各损害类型的影响因素颇多,每种因素在损害中具体发挥多大作用是人们研究的重点和难点。以往进行储层损害预测都采用裸权或者是专家咨询法赋权,前者忽略了各特征属性对损害结果的相对贡献度不同这一事实,后者过分依赖专家的主观判断和经验,其结果有时难以令人信服。层次分析法将组成复杂问题的多个元素权重的整体判断转变为对这些元素进行"两两比较",然后对这些元素的整体权重进行排序判断,最后确立各元素的权重。以水敏为例,采用层次分析法对其影响因素进行了权值赋予,最后用模式识别检验了因素赋权对准确率的影响。结果表明:层次分析法将决策者的主观判断及客观推理紧密地联系起来,可以较合理地对各损害因素进行赋权;赋权后比赋权前准确率平均提高了8.96%,个别准确率提高明显,甚至达到了17.08%,这将会对储层损害快速预测方法的建立起到巨大的推动作用。
The rapid prediction model of reservoir damage can timely and accurately provide potential types and extents of damaged reservoirs,which can give better service to reservoir protection.However,there are too many factors that cause reservoir damage,and what role each factor individually plays is important and difficult to be predicted.In the past,the naked power or consulting experts were commonly trusted in establishing a model to predict reservoir damage,however,the former is apt to ignore the fact that each characteristic plays different roles in causing damage to reservoirs while the latter often suffers from overreliance on experts＇ subjective judgments and experience that are sometimes unreliable.The analytical hierarchy process（AHP） can transform the overall judgment on multi-element weights that constitute a complicated problem into a pairwise comparison between these elements,then sort,judge and finally determine the overall weight of each element.The present paper took water sensitivity as an example to evaluate the weight of influential factors by means of the AHP and used the pattern recognition to test the influence of factor weighting on accuracy.The result showed that the AHP could connect subjective judgments with objective reasoning closely,the weight of each damage factor could be determined more reasonably,and the individual accuracy rate could be elevated by 8.96% or even 17.08% after weighting,which,therefore,would play a huge boosting role in establishing a rapid forecast method to predict reservoir damage.
出处
《石油学报》
EI
CAS
CSCD
北大核心
2011年第6期1037-1041,共5页
Acta Petrolei Sinica
基金
国家杰出青年科学基金项目(50925414)
国家科技重大专项(2011ZX05009-005)联合资助
关键词
层次分析法
权值确定
特征提取
敏感性预测
模式识别
analytical hierarchy process（AHP）
weight determination
feature extraction
sensitivity prediction
pattern recognition
