用扩散模型快速生成地下流体流动模拟，速度提升100倍
LAViG-FLOW: Latent Autoregressive Video Generation for Fluid Flow Simulations
- 基于潜空间自回归的视频生成框架，联合建模压力与饱和度演化
- 在CO2封存数据集上生成结果时序一致，推理速度比传统方法快100倍
- 适合需要大量模拟的地质封存与地热开发场景
地下多相流体流动建模与预测对地质碳封存（GCS）和地热生产等应用至关重要，关乎运行性能与长期安全。尽管高保真多相模拟器广泛应用，但多次前向求解用于反演和不确定性量化时成本过高。为此，我们提出LAViG-FLOW，一种潜空间自回归视频生成扩散框架，显式学习饱和度与压力场的耦合演化。每个状态变量由专用2D自编码器压缩，视频扩散变换器（VDiT）建模其时间上的联合分布。模型先在给定时间窗口内训练以学习耦合关系，再通过自回归微调实现超出观测时间窗的外推。在开源的CO2封存数据集上评估，生成的饱和度与压力场保持时序一致性，且推理速度比传统数值求解器快两个数量级。
原文摘要 · Abstract (English)
Modeling and forecasting subsurface multiphase fluid flow fields underpin applications ranging from geological CO2 sequestration (GCS) operations to geothermal production. This is essential for ensuring both operational performance and long-term safety. While high fidelity multiphase simulators are widely used for this purpose, they become prohibitively expensive once many forward runs are required for inversion purposes and to quantify uncertainty. To tackle this challenge, we propose LAViG-FLOW, a latent autoregressive video generation diffusion framework that explicitly learns the coupled evolution of saturation and pressure fields. Each state variable is compressed by a dedicated 2D autoencoder, and a Video Diffusion Transformer (VDiT) models their coupled distribution across time. We first train the model on a given time horizon to learn their coupled relationship and then fine-tune it autoregressively so it can extrapolate beyond the observed time window. Evaluated on an open-source CO2 sequestration dataset, LAViG-FLOW generates saturation and pressure fields that stay consistent across time while running two orders of magnitude faster than traditional numerical solvers.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。