论文部分内容阅读
径向基(RBF)神经网络法具有网络结构简单、逼近能力强和学习速度快等优点,已成为最具发展潜力的储层敏感性智能预测方法之一,但在实际应用中仍存在泛化能力不强、网络训练不收敛等问题。通过在输入层中引入补充节点,对网络拓扑结构进行优化,有效地提高了RBF神经网络的逼近精度和泛化能力。在确定储层敏感性主要影响因素的基础上,通过对径向基函数散布常数的优选,进一步优化了RBF神经网络的性能。采用所收集的胜利、辽河、大港及江苏油田共125组数据,进行了神经网络训练和预测检验,优化了RBF神经网络,并在储层敏感性预测方面进行了应用。结果表明,对于训练集内的样本,预测的平均准确率均大于93.79%,且预测值与实验值的相关系数均大于0.995;对于训练集外的样本,预测的平均准确率大于91.59%,预测值与实验值的相关系数大于0.994,实现了对储层敏感性的准确、定量预测。
Radial Basis Function (RBF) neural network method has the advantages of simple network structure, strong approximation ability and fast learning speed, which has become one of the most promising reservoir sensitivity intelligent prediction methods, but there are still some generalizations Ability is not strong, network training does not converge and so on. By introducing complementary nodes into the input layer, the topology of the network is optimized to effectively improve the approximation accuracy and generalization ability of the RBF neural network. On the basis of determining the main influencing factors of reservoir sensitivity, the performance of RBF neural network is further optimized by optimizing the distribution constant of radial basis function. A total of 125 sets of data collected from Shengli, Liaohe, Dagang and Jiangsu Oilfields were used for neural network training and prediction tests. The RBF neural network was optimized and applied to the prediction of reservoir sensitivity. The results show that the average accuracy of prediction is greater than 93.79% for the samples in the training set, and the correlation coefficient between the predicted value and the experimental value is greater than 0.995. For the samples outside the training set, the average prediction accuracy is greater than 91.59% The correlation coefficient between the value and the experimental value is greater than 0.994, which realizes accurate and quantitative prediction of reservoir sensitivity.