GeoLog-Agent 用 LLM 地质先验重建缺失测井曲线
Well-Log Reconstruction with Pre-trained Geological Knowledge: A Knowledge-Conditioned Large-Language-Model Generative Agent
石油勘探里测井数据常有缺失段,传统回归模型缺乏地质知识补不准。这篇把 LLM 当地质先验接进生成流程,北海数据上 R² 从 0.84 拉到 0.93。做地质数据分析或科研工程交叉方向的人可以看看这个范式。
挪威北海基准上,GeoLog-Agent 在测井缺失段重建任务中取得平均 R² 0.9307、峰值 R² 0.9713,比九个基线的平均 R² 0.8387 高出 0.0920,平均 RMSE 降低 30.97%,对最强基线的 R² 领先 0.0274–0.0498,RMSE 低 16.25%–24.05%。
技术核心是把预训练 LLM 作为地质先验来源:一个通用 Geo-Harness 将 LLM 的地层推理转化为以地层为锚点的岩石物理轨迹,配合特征级线性调制(FiLM)与条件对抗约束来约束生成过程,解决缺失测井反演的病态问题。消融实验验证了地质先验推理、多尺度空洞生成和条件对抗三个模块的协同作用。
研究者还把这套耦合机制迁移到六个常规深度学习模型上,使它们的 R² 提升 0.0162–0.0789 且保持跨井稳定性,说明基础模型的领域推理能力可以泛化为对生成式重建任务的约束手段。
Well-Log Reconstruction with Pre-trained Geological Knowledge: A Knowledge-Conditioned Large-Language-Model Generative Agent
Missing well-log intervals make subsurface reconstruction ill-posed. Existing numerical regression models lack geological prior knowledge to constrain ambiguous inverse mappings. We propose GeoLog-Agent, a knowledge-conditioned generative agent using pre-trained large language models (LLMs) as a source of geological priors. A universal Geo-Harness translates LLM stratigraphic reasoning into formation-anchored petrophysical trajectories, feature-wise linear modulation, and conditional adversarial constraints. On the Norwegian North Sea benchmark, GeoLog-Agent achieves a mean $R^2$ of 0.9307 and a peak $R^2$ of 0.9713. It exceeds the nine-baseline average $R^2$ of 0.8387 by 0.0920, with a 30.97% average RMSE reduction, and exceeds the strongest baseline by 0.0274--0.0498 in $R^2$, with 16.25%--24.05% lower RMSE. Ablations confirm the effectiveness and synergy of geological prior reasoning, multi-scale dilated generation, and conditional adversarial modules. Transferring the coupling mechanism to six conventional deep learning models improves their $R^2$ by 0.0162--0.0789 with cross-well stability. These results demonstrate how foundation-model domain reasoning can constrain deep generative reconstruction of missing petrophysical measurements.
This report is categorized under AITOP's "Papers" desk. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include GeoLog-Agent, LLM, generative models.
Source attribution: curated from arXiv cs.AI (@Yongan Zhang, Yitao Yu, Yizheng Li, An Gong, Youzhuang Sun).