摘要
根据地层可钻性时间序列特征,应用支持向量机理论,提出了一种对钻头下部未钻开地层的可钻性进行预测的地层可钻性时序支持向量机预测方法,并建立了基于支持向量机的地层可钻性时序预测模型。应用该方法对长庆油田富古1井的地层可钻性进行了预测。将该预测结果与BP神经网络方法的预测结果进行对比分析的结果表明,该方法优于BP神经网络方法,具有预测精度高、推广预测能力强等优点。
The evolvement characters of formation driliability time series were analyzed. A new method for predicting driliability of un drilled formation under the bit was proposed according to the theory of support vector machine. A prediction model for formation driliability time series was given. This method was applied to predict the formation driliability of Fugu 1 Well in Changqing Oilfield. The comparison of the prediction results with the results of BP neural network indicates that this method is better than BP neural network and has the advantages of high prediction accuracy and excellent generalization. This method is suitable for formation driliability prediction before drilling.
出处
《石油学报》
EI
CAS
CSCD
北大核心
2006年第1期97-100,共4页
Acta Petrolei Sinica
基金
国家自然科学基金重点项目(No.50234030)
国家自然科学基金重大研究计划项目(No.90410006)联合资助
关键词
地层
岩石可钻性
时间序列
支持向量机
神经网络
预测模型
formation
rock driliability
time series
support vector machine
neural network
prediction model
