论文部分内容阅读
论述了油田区块开发前期和中期的油气预测的方法及实际应用问题．主要利用三维地震资料提取地震属性参数，并采用模糊神经网络技术，即将模糊概念和ＢＰ神经网络结合一体进行预测．为使ＢＰ神经网络尽快收敛，减小振荡，增强网络的可靠性，还提出了两种主要修正ＢＰ网络的方法，其一为动态调节学习率；其二为动态调整权系数参数．通过实例分析认为：该方法的精度和可靠性较高，其预测结果为油田开发区块增产上储提供了较可靠的依据．
The method and practical application of oil and gas prediction in the early and middle stages of oilfield development are discussed. The seismic attribute parameters are mainly extracted by using 3D seismic data, and the fuzzy neural network technology is adopted, that is, the fuzzy concept and the BP neural network are combined to predict the seismic attributes. In order to make the BP neural network converge as soon as possible, reduce the oscillation and enhance the reliability of the network, two main methods to correct the BP network are also proposed. One is to dynamically adjust the learning rate; the other is to dynamically adjust the weight coefficient parameters. The case study shows that the method has high accuracy and reliability, and its prediction result provides a reliable basis for the increase of reserves in the oilfield development block.