摘要
测井解释过程中的岩性识别实质是多个指标数据的模式识别问题。常规测井解释方法很难表征储层的真实特性。提出一种基于主成分分析的最小二乘支持向量机的岩性识别预测模型(PCA-LSSVM)。介绍了主成分分析法和最小二乘支持向量机原理。通过主成分分析方法对测井数据进行分析并提取影响岩性识别的主要因素,依据分析结果建立基于最小二乘支持向量分类机的岩性识别模型。云南陆良盆地3口井的117个地层的识别结果与实际取心资料的符合率达到92.5%。应用表明,将主成分分析结合最小二乘支持向量机进行岩性识别,简化了网络结构,具有更快的运算速度和准确率,是一种值得推广使用的方法。
Oil and gas reservoir recognition is virtually pattern recognition in the process of well log interpretation with several indexes. It＇s difficult to show the reservoir characteristics by con ventional log interpretation methods. Proposed are a method of identifying lithology based on principal component analysis and least square support vector machine （PCA-LSSVM）. Introduced are the theory of principal component analysis and least square support vector machine. The extracted principal components from log data by use of principal componenl analysis method are the main factors that influence the lithology identification. Then the identifying lithology model of least square support vector machine classifier is got by the analysis results, The coincidence rate of lithology identification and practical coring data for 117 layers of 3 wells in Luliang Basin reaches 92, 5%. The results indicate that the least square support vector machine identif ying lithology model combining with principal component analysis method could get faster speed and higher accuracy than that of nature network and support vector machine, it is worthy of further study and widely use.
出处
《测井技术》
CAS
CSCD
北大核心
2009年第5期425-429,共5页
Well Logging Technology
基金
四川省教育厅重点项目(07ZA143)资助
关键词
测井解释
岩性识别
主成分分析
最小二乘支持向量机
累积方差
log interpretation, lithology identification, principal component analysis, teast squaresupport vector machine, cumulative contribution
