论文部分内容阅读
孔隙度是油气藏描述的一个重要参数。基于双相介质中地震波传播理论，论述了地震孔隙度预测原理。考虑到地震孔隙度预测的复杂性与ＢＰ网络函数逼近需要利用全体样本的信息、学习效率低（不适于用来优选地震特征）等不足，提出采用完全利用样本信息（ＣＵＳＩ）的网络做孔隙度预测。该方法利用ＣＵＳＩ网络的局部逼近功能，依据井孔数据与井旁地震数据建立地震特征与孔隙度的函数关系来预测孔隙度。在此基础上还提出了ＣＵＳＩ网络孔隙度预测中的地震特征优化原理和基于遗传算法的地震特征优化方法。实际应用结果表明：此方法明显改善了地震孔隙度的预测精度，具有实用价值。
Porosity is an important parameter described by reservoirs. Based on the theory of seismic wave propagation in two-phase media, the principle of seismic porosity prediction is discussed. Considering the complexity of the prediction of seismic porosity and the approximation of the function of BP network, it is necessary to make use of the information of the whole sample, the low learning efficiency (not suitable for the selection of the seismic characteristics), and the use of the network of complete sample information (CUSI) prediction. The method uses the local approximation function of CUSI network to predict the porosity based on the correlation between the seismic data and the seismic data near the well as a function of porosity. On this basis, the principle of seismic feature optimization in CUSI network porosity prediction and the method of seismic feature optimization based on genetic algorithm are also proposed. The practical application shows that this method can significantly improve the prediction accuracy of seismic porosity and is of practical value.