论文部分内容阅读
根据油气层综合解释的特点，提出了单井油气层综合解释的人工神经网络方法。采用的神经网络模型为包含２８个输入神经元、２６个隐含神经元和４个输出神经元的３层网络。选择了松辽盆地北部大庆长垣以西地区９１个试油层段的资料对网络进行了训练，通过对已知样品的交叉验证，训练后的神经网络识别符合率为０．９３４，证明人工神经网络是一种有效的油气层综合解释方法。
According to the characteristics of comprehensive interpretation of oil and gas layers, an artificial neural network method for comprehensive interpretation of single well is proposed. The neural network model used is a 3-layer network containing 28 input neurons, 26 implicit neurons and 4 output neurons. Based on the data from 91 oil reservoirs west of Daqing Changyuan in the north of Songliao Basin, the network was trained. The cross-validation of known samples shows that the coincidence rate of trained neural network is 0.934, which proves that artificial neural network Is an effective method of comprehensive interpretation of the reservoir.