報(bào)告題目：A glimpse of the researches and applications of deep learning for digital rock physics

報(bào)告人：蔣澤云（Heriot-Watt大學(xué)，教授）

報(bào)告時(shí)間：2021年6月11日16:00-18:00

報(bào)告地點(diǎn)：騰訊會(huì)議ID：259 474 071

摘要:The advent of deep learning (DL) marked a milestone in the real-life applicability, as now very complex problems can be solved with unprecedented accuracy. DL generally require little explicit prior knowledge and are distinctively efficient in extracting complicated patterns. These capabilities turn them into feasible candidates for replacing and/or assisting conventional time-consuming and computationally expensive experiments.

This talk aims to show how the power of deep learning can be harnessed to both estimate porous-media properties and develop new insights. The main objectives are: (1) provide a general overview of how DLs have already been used in terms of single/multi-phase flow; (2) demonstrate the potentials of DL in digital rock physics through case studies; (3) discuss DL-based approaches to explore the physics of the porous media.

First, the relevant body of research is considered so that advancements, gaps and potentials can be identified. Then, an implementation map is laid out, encompassing the simplest to most comprehensive applications. Secondly, several cases are presented to show-case its ability. Thirdly, future research is briefly discussed. It is proposed that to develop reliable multi-phase predictors, large databases must be synthesized by collecting, resampling, augmenting, and grouping images and the corresponding properties. Consequently, deep neural networks can be trained for various rock types and processes. Singular or ensembles of DL networks may either be used to make predictions or to serve as the base to be customized for other applications, i.e., transfer learning. Final models can be put to ultimate real-life testing by comparing against experimental data, e.g., phase distributions from synchrotron imaging. Rather than creating mere black-box estimators, one must strive to understand how the networks extract information and link relationships, by looking at layer architectures, weights and other elements. The goal should be to gain insights into various flow functions and the physics of certain flow behaviours. Furthermore, since trained models are very fast to run, they make perfect assets for such tasks as sensitivity/uncertainty analysis and back-calculation of input features, for instance, to see what wettability distribution can result in a specific flow parameter.

報(bào)告人簡(jiǎn)介：蔣澤云教授從2004年以后主要從事孔隙介質(zhì)（如巖石、土壤等）異質(zhì)多尺度結(jié)構(gòu)分析和流體滲流模型研究,在Water Resource research, Transport in Porous Media, Fuel等期刊和國(guó)際會(huì)議上發(fā)表論文40多篇，在國(guó)內(nèi)外參與并主導(dǎo)若干重大科研項(xiàng)目。主要從事微觀空隙結(jié)構(gòu)分析及網(wǎng)絡(luò)滲流模型的研究,獨(dú)立開發(fā)軟件系統(tǒng)PAT–Pore Analysis Tools（這一軟件在學(xué)術(shù)和工業(yè)界得到廣泛的使用）。擅長(zhǎng)于巖石微觀圖像幾何拓?fù)鋵傩缘姆治?并建立多尺度孔隙模型和實(shí)施數(shù)值模擬,建立其微觀結(jié)構(gòu)（如孔隙度、孔尺寸、形狀、連通性、孔壁粗糙度等）與宏觀流體屬性（如滲透率、毛細(xì)管壓力、性、相對(duì)滲透率、電阻率等）間的理論或經(jīng)驗(yàn)公式。

熱烈歡迎全校師生參加學(xué)術(shù)交流！

舉辦單位：科研處、理學(xué)院、人工智能研究院