论文部分内容阅读
随着石油勘探领域的不断扩大,含油性识别的研究对象也越来越复杂,传统的基于单一硬计算或软计算的方法在含油性识别中面临着严峻挑战.首先提出了软计算与硬计算融合的4种模式,然后运用GA-FCM对含油性的测井属性进行约简,将约简后的测井属性结合软计算与硬计算融合的分离模式对某油田Oilsk81,Oilsk83,Oilsk85三口井进行含油性模式识别,去掉出错率较高的样本,达到样本约简的目的;最后利用判别分析法对约简后的样本集进行检验分析.实验表明:第一,在这几个油区可以用声波时差和含油饱和度两个测井属性进行含油性识别;第二,将出错率高的样本进行约简可以提高样本集识别的正确率.
With the continuous expansion of the field of oil exploration, the research object of oil-bearing identification is more and more complex, and the traditional methods based on single hard computing or soft computing are facing severe challenges in oil-containing identification.Firstly, soft computing and hard computing Then we use GA-FCM to reduce oil-bearing logging properties, and combine the reduced logging properties with the soft-hard and hard-data separation modes. In the three oil wells Oilsk81, Oilsk83 and Oilsk85 The oil mist pattern recognition is carried out and the sample with higher error rate is removed to achieve the purpose of sample reduction.Finally, the discriminant analysis method is used to test and analyze the reduced sample set.The experiment shows that: firstly, Second, to reduce the samples with high error rate, the correctness of the sample set recognition can be improved.