摘要
支持向量机(SVM)算法是特别适合于用有限已知样本训练建模,进而预报未知样本属性的模式识别新算法。笔者尝试将Vapnik提出的支持向量机算法用于水淹层测井识别。总结了P油田水淹层的声波时差、自然电位、深感应电阻率、中感应电阻率及密度测井曲线与水淹程度的对应关系,建立了基于支持向量分类机的识别模型,并将上述参数作为训练样本的输入,油气特征作为训练样本的输出,对支持向量机进行训练。对于P油田水淹层的实际预测结果表明:支持向量机可以成为一种用于水淹层识别的有效工具。
The support vector machine proposed by Vapnik is a newly-developed technique for data processing. It is suitable for the data processing based on a finite number of training samples, with special technique for restricting overfitting. In this paper, the support vector classification technique was used to make modeling on the relationships between the acoustic time, SP, deep induction resistivity, medium induction resistivity, density and water flood grade, with these parameters serving as input of the training samples and the character of the oil and gas as the output. This technique was used in the P oilfield, which shows that SVM can yield efficient modeling results.
出处
《物探与化探》
CAS
CSCD
2008年第6期652-655,共4页
Geophysical and Geochemical Exploration
关键词
水淹层
测井识别
数学模型
模式识别
支持向量机
Flooding formation
Well logging recognize
Mathematic modeling
pattern recognition
Support vector machine