碎屑岩油气水层神经网络解释模型

该文针对地质录井油气水层综合解释所用传统方法存在的问题,介绍一种基于人工神经网络的模式识别新方法。采用的神经网络模型为四层BP网络。输入层有14个神经元,第一隐含层25个神经元,第二隐含层14个神经元,输出层有4个神经元。选用了辽河油区400个碎屑岩层样本对网络进行训练,训练后网络识别率达到96.5%。说明人工神经网络是一种有效的综合解释油气水层的新方法。

The Neural Net Interpretation Model For Oil Gas And Water Beds In Clastic Reservior

The paper introduces a way based on the simulation identification with the manual neural net in order to deal with the problems using the traditional method.The neural net used consists of layers of BP nets,which includes the input layer composed of 14 neural cells,the first implied layer composed of 25 neural cells,the second composed of 14 neural cells and the output layer composed of 4 neural cells.The net has been trained through 400 clastic samples in Liaohe Oilfield and its identification rate reaches 96.5 percent.So this shows the neural net is an effective method in the comprehensive interpretation for the oil,gas and water beds.