| 标题 |
Well‑log super‑resolution reconstruction with multiscale feature‑guided U‑Net for thin‑bed reservoir characterization |
| 网址 | |
| DOI |
10.1016/j.petrol.2023.110789
doi
|
| 其它 |
**Title**:Well‑log super‑resolution reconstruction with multiscale feature‑guided U‑Net for thin‑bed reservoir characterization **Journal**:*Journal of Petroleum Science and Engineering* (JCR Q1,石油工程一区) **DOI**:10.1016/j.petrol.2023.110789 **永久链接**:[https://doi.org/10.1016/j.petrol.2023.110789](https://doi.org/10.1016/j.petrol.2023.110789) **核心**:专门针对常规测井曲线纵向 2 倍上采样(0.25 m→0.125 m);改进 U‑Net,增加多尺度边缘损失;解决深度学习超分辨把薄层峰谷抹平、幅值偏移问题;采用 GR、RHOB、NPHI 实井数据;对比双三次插值、SRCNN、普通 UNet;高频突变位置保真度作为核心评价指标。 |
| 求助人 | |
| 下载 |
PDF的下载单位、IP信息已删除
(2025-6-4)