径向基函数神经网络在启动压力梯度预测中的应用

Application of Radial Basic Function Neural Network on the Prediction of Starting Pressure Gradient

启动压力梯度直接影响着低渗透油田的开采量以及油藏压力的预测精度,它与流体粘度、密度、渗透率、孔隙度等影响因素呈非线性关系。而人工神经网络具有表示任意非线性关系和学习的能力,是解决复杂非线性、不确定性和时变性问题的新思想和新方法。利用径向基函数(RBF)人工神经网络对启动压力梯度进行预测,并结合岩心的启动压力梯度的实际测定结果进行研究,结果表明:RBF人工神经网络是一种较为有效的预测方法,具有较高的精度,该方法可以为低渗油田的开发提供可靠的基础数据,节省了人力、物力。

Starting pressure gradient directly affects the recovery rate of low permeable oilfield and the prediction accuracy of reservoir pressure, it bears nonlinear relation to some factors like fluid viscosity, density, permeability and porosity . While the artificial neural network has the ability to indicate the random nonlinear relation and the ability for study, it has become a new idea and a new method to solve the complex nonlinear, indefinite and time - varying problems . By adopting radial basic function(RBF) neural network to predict the starting pressure gradient, and by combining the real measured results of the starting pressure gradient, the study results show that RBF artificial neural network is a kind of effective prediction method with high accuracy, this method can provide the reliable basic data for the development of low permeable oilfield, saving a great deal of labor power and costs.