摘要
岩性识别一直是储层测井解释的关键问题和难点之一。针对常规测井岩性识别准确率不高的状况,在分析测井资料的基础上,以Matlab为平台研究了基于主成分分析的PCA-BP神经网络,并以济阳坳陷非常规储层实际测井资料为样本,通过设计算法步骤进行了实验仿真。由仿真结果得出非常规储层岩性识别率为95.8%,高于BP神经网络,PCA-BP神经网络有效提高了识别率和运行速度。经过对济阳坳陷钻井的岩性识别表明,该岩性识别方法可行并具有实用价值。
Lithology identification has been the key and difficult point of reservoir logging interpretation.Considering the low accuracy of conventional lithology identification methods,the BP neural network based on improved principal component analysis（PCA）was studied on the basis of logging data analysis and with Matlab as the platform.The actual logging data of unconventional reservoir in Jiyang sag was taken as sample,on which experiment simulation was performed by designing algorithm.The simulation results show that with the unconventional reservoir lithology identification rate of 95.8%,which is higher than BP neural network,PCA-BP neural network is effective to improve the identification rate and speed.The logging lithology identification in Jiyang sag proves that this lithology identification method is feasible and has practical values.
出处
《山东科技大学学报（自然科学版）》
CAS
2016年第5期9-16,共8页
Journal of Shandong University of Science and Technology(Natural Science)
基金
国家自然科学基金项目(41472092)
关键词
非常规储层
济阳坳陷
岩性识别
主成分分析
BP神经网络
unconventional reservoirs
Jiyang sag
lithology identification
principal component analysis
BP neural network
