用灰色支持向量机进行储层油气、水模式识别  

Pattern Recognition of Oil and Gas Reservoirs Based on Grey Support Vector Machine

作　　者：刘冬娥[1] 黄婧芝[1] 吴国平[1] 

机构地区：[1]中国地质大学(武汉)机械与电子信息学院,湖北武汉430074

出　　处：《石油天然气学报》2009年第4期261-264,共4页Journal of Oil and Gas Technology

基　　金：国家自然科学基金项目(40674069)

摘　　要：在石油天然气开发中,储层油气、水判别属于典型的模式识别问题。为了克服传统的学习方法存在的过于简单、对经验较强的依赖、涉及人为因素多、易陷入局部最小值、存在过学习问题等缺陷,同时也为了适应油气储层样本属性特征差异小,属性特征信号信噪比不够高的特点,在研究灰色关联和支持向量机原理的基础上提出灰色支持向量机模型,用于储层油气、水模式识别。与单纯的支持向量机识别方法相比较该模型判别油气、水属性的正确率高。研究结论可用于油气勘探储层精细评价和油气勘探二次开发。Oil and gas reservoir recognition was a typical problem of pattern recognition in oil and gas exploration.Traditional learning method such as linear classifier,Bayesian classifier,fuzzy pattern recognition and BP algorithm had their shortcomings such as strong dependence on experience,too many human factors involved,easily limited by local minimum and so on.Difference of attributes of oil and gas reservoir samples was small and Signal-to-Noise was low.To overcome those shortcomings and fit the peculiarity of the oil and gas reservoir samples,the model of grey support vector machine is proposed based on theory of grey relation and support vector.This model is used to recognize oil and gas reservoirs.In practical application and research,logging data of ST well in some oil fields in Northwest of China were used.The data includes SP,RFOC,GR,AC,which are used to compose 198×4 sample space.The recognition rate of this algorithm is 97.56%,while it is only 85.37% of support vector machine,improving correct rate is 12.19%.The conclusion can be used for fine reservoir evaluation and secondary development of oil and gas exploration.

关 键 词：灰色关联 支持向量机 油气 水 模式识别 

分 类 号：P618.13[天文地球—矿床学] N941.5[天文地球—地质学]