摘要
目前剩余油的研究是世界性的热点问题。论文利用神经网络任意逼近非线性及自学习的能力,将BP(Back Propagation)神经网络算法应用于剩余油孔隙度、渗透率及饱和度三个参数的预测研究。以大庆油田某油区数据为例,采集该油区内若干口油井的三维地理位置信息及孔渗饱参数,将其中的三维地理位置信息数据作为神经网络的输入参数,孔渗饱三参数作为神经网络的输出参数。处理输出数据时,以是否具有经济开采价值为标准,对数据进行分类,利用MATLAB软件进行编程仿真,该方法对已知数据信息种类要求不高,简单易行。仿真表明,论文给出的方法实现了通过直接输入某油区三维地理位置来预测该位置处孔渗饱参数情况,预测结果准确率高,为实际油田生产提供了很好的辅助作用。
At present the research of remaining oil is the world focus.By use of neural networks'(NN) ability of approximating nonlinear and self-learning,BP algorithm is applied to predict porosity,permeability and saturation of remaining oil.Taking data of an oil region wells for example,collect the three-dimension location information and porosity,permeability and saturation parameters of the oil region in Daqing is as the NN input.The three parameters are as the NN output.Classify the output data based on exploitation economic value and program in MATLAB.The given prediction method doesn't require much more known information types,so the method is easy to realize.The simulation result shows that the method given in this paper accurately predicts three parameters in an oil region- porosity,permeability and saturation based on known three-dimension location.The prediction result accuracy is high enough to help enhance oil recovery.
出处
《内蒙古石油化工》
CAS
2013年第23期47-50,共4页
Inner Mongolia Petrochemical Industry
关键词
BP神经网络
剩余油
预测
孔隙度
渗透率
饱和度
BP neural networks
Remaining oil
Prediction
Porosity
Permeability
Saturation