摘要
顺9井区位于塔里木盆地塔中隆起带中顺托果勒低隆构造单元顺1三维工区,研究区有效砂岩储层厚度较薄,平面分布差异大,利用波阻抗反演及谱白化方法预测储层分布效果不理想。而统计分析频谱系列属性与薄砂岩层厚度关系密切,因此,利用BP神经网络作为分频信息的融合手段,以此来反演有效砂岩的储层厚度。预测结果与钻井有效砂岩储层厚度符合度较高,证明该方法有效可靠。
Wellblock Shun 9was located in Shun 1 3D work area in the tectonic unit of Shuntuoguole low uplift of Tazhong Uplift in Tarim Basin,the thickness of effective sandstone reservoir in the studied area was thin,plane distribution difference was big,wave impedance inversion and spectral whitening methods were used to predict reservoir distribution,its effect was not ideal.But statistical analysis of frequency spectrum series properties was closely related to the thickness of thin layer sandstone,thus,BP neural network was used as fusion method of separate frequency information for the inversion of effective sandstone reservoir thickness.Prediction results and drilling effective thickness of sandstone reservoir accords with higher degrees,it proves that this method is effective and reliable.
出处
《石油天然气学报》
CAS
CSCD
2014年第1期45-49,5-6,共5页
Journal of Oil and Gas Technology
关键词
地震属性
属性优化
储层预测
神经网络
分频
seismic attribute
attribute optimization
reservoir prediction
artificial neural network
separate frequency