人工神经网络系统在储层评价中的应用  

The Application of Artificial Nerve Network System for Reservoir Evaluation

作　　者：姜延武[1] 严桂林[2] 孔郁琪 徐东旭[3] 王发展[2] 

机构地区：[1]吉林油田地质录井公司研究所 [2]吉林石油集团地质录井公司 [3]中油吉林油田分公司新民采油厂

出　　处：《录井技术》2001年第2期8-12,共5页

摘　　要：该文利用人工神经网络原理及误差反向传播算法,将荧光录井、气测录井参数量化及标准化处理,建立以五参数为输入模式样本的人工神经网络解释模型,对储层特征进行描述,在吉林探区以试油层指标进行网络训练,四类层的回代检验率总体达到95%,对未试油层训练解释与反馈试油结果对比,正判率为83.3%,适合现场随钻快速评价储层的需要。This paper uses artificial network′s principle and error backpropagation algorithm to make the fluorescence logging and gas logging parameters quantitative and standized,so the artificial nerve network interpretation model has been set up based on five parameters′ input sample.The model describes the reservoir characteristics.During the network training according to the tested reservoir index in Jilin exploration area,total verified check rate for four classes of layer is 95 percent,while compared with the untested oil layer′s trainning interpretation and feedback oil test results,the rate for positive judgment is 83.3 percent ,which meets the requirements of field rapid evaluation while drilling.

关 键 词：人工神经网络 储层评价 气测井 荧光录井 

分 类 号：TE155[石油与天然气工程—油气勘探] TP183[自动化与计算机技术—控制理论与控制工程]