论文部分内容阅读
常规测井解释孔隙度的方法是运用线性响应方程求解,或用统计方法建立测井曲线与孔隙度之间的统计关系模型求解,但这些方法在面对越来越复杂的地质条件和非均质性的研究对象(如碳酸盐岩地层)时,所得出的结果与地层的实际数值存在着较大误差。利用对碳酸盐岩地层孔隙度敏感的测井值建立样本,利用BP神经网络预测碳酸盐岩地层孔隙度,预测孔隙度与岩心孔隙度有良好的符合关系,孔隙度逐点对应的绝对误差普遍小于1·0孔隙度单位。
Conventional well logging methods to interpret porosity are by using a linear response equation or by statistical methods to establish statistical relationships between well logs and porosity. However, in the face of increasingly complex geological conditions and non-homogeneous Qualitative research objects (such as carbonate rock formation), the results obtained and the actual formation of the actual value there is a big error. A sample was established by using the logging values that are sensitive to carbonate formation porosity. The BP neural network was used to predict the porosity of the carbonate rock formation. The porosity was predicted to have a good correspondence with the core porosity, and the porosity was correspondingly The error is generally less than 1.0 porosity unit.