Some applications of Deep Learning techniques to fluid dynamic solutions
Danesh K. Tafti
William S. Cross of Engineering
Dept. Mechanical Engineering
Virginia Tech, Blacksburg, USA
講座時(shí)間：
2023年9月10日 上午10:00-11:00
講座地點(diǎn)：
三號(hào)實(shí)驗(yàn)樓307
講座內(nèi)容簡(jiǎn)介：
考慮到計(jì)算流體動(dòng)力學(xué) (CFD) 解決方案的復(fù)雜、昂貴及不確定性，人工智能 (AI) 和深度學(xué)習(xí) (DL) 方法在計(jì)算成本及結(jié)果準(zhǔn)確度上的優(yōu)勢(shì)讓它在流體力學(xué)領(lǐng)域獲得了越來(lái)越多的重視。該法可通過(guò)更好地參數(shù)化現(xiàn)有模型來(lái)實(shí)現(xiàn)，因而其常被應(yīng)用在湍流建模、新模型開(kāi)發(fā)或者通過(guò)加速傳統(tǒng)求解算法或開(kāi)發(fā)降階代理模型來(lái)降低 CFD 成本等方面。然而，納維-斯托克斯方程中包含的復(fù)雜非線性物理特征與高昂的生成訓(xùn)練數(shù)據(jù)的成本是這些方法推廣過(guò)程中亟待解決的問(wèn)題與挑戰(zhàn)。該講座將探討深度學(xué)習(xí)在流場(chǎng)預(yù)測(cè)與偏向工程導(dǎo)向方面應(yīng)用，內(nèi)容包含兩方面的案例研究：第一個(gè)案例研究了隨機(jī)分布的柱狀顆粒集合中隨時(shí)間變化的混沌流場(chǎng)的未來(lái)狀態(tài)預(yù)測(cè)；第二個(gè)案例研究了不同堆積密度和雷諾數(shù)下隨機(jī)分布的長(zhǎng)橢球顆粒集合中的穩(wěn)定流場(chǎng)的預(yù)測(cè)。此外，研究還對(duì)通過(guò)預(yù)測(cè)流場(chǎng)計(jì)算出的顆粒受力等物理量的準(zhǔn)確性進(jìn)行了評(píng)估，證實(shí)了當(dāng)前模型的準(zhǔn)確性與可靠性。
Introduction:
Computational Fluid Dynamics (CFD) solutions are complex, expensive, and uncertain. Artificial Intelligence (AI) and Deep Learning (DL) methods have the potential to give accurate results at much less computational cost. This can be through better parameterization of existing models, e.g. turbulence modeling, or through the development of new models where none were possible, or through reducing the cost of CFD by accelerating conventional solution algorithms or by the development of reduced-order surrogate models. However the complex non-linear physics embedded in the Navier-Stokes equations and the cost of generating training data (data paucity) are some of the challenges that impede the generalizability of these methods. The seminar will explore the prediction of flow fields and downstream engineering tasks such as determining forces acting on embedded objects through the use of DL techniques. In the first case study, the future state prediction of a time-dependent chaotic flow field in a random array of cylinders is investigated. In the second case study the prediction of steady flow fields in different random assemblies of prolate ellipsoids under different packing densities and Reynolds numbers is investigated. In both case studies, the accuracy with which engineering quantities such as drag forces can be found using the DL predicted flow fields is also evaluated.
主講人簡(jiǎn)介：
Danesh. K. Tafti 教授擁有三十余年計(jì)算流體力學(xué)相關(guān)的研究與工作經(jīng)驗(yàn)，研究方向涵蓋了大渦模擬及高性能并行運(yùn)算算法開(kāi)發(fā)、撲翼飛行的空氣動(dòng)力學(xué)分析、顆粒-流體兩相流及泥沙輸移高精度仿真等方向，共發(fā)表論文二百七十余篇，其中有學(xué)術(shù)期刊論文一百四十余篇，總引用數(shù)達(dá)7179次（來(lái)源：谷歌學(xué)術(shù)，2023年9月）。Tafti教授曾任《ASME J. Heat Transfer》副主編，目前為《International Journal of Heat and Fluid Flow》、《Journal of Applied and Computational Mechanics》及《International Journal of Rotating Machinery》期刊編委。Tafti 教授多年來(lái)一直保持著與美國(guó)國(guó)家能源實(shí)驗(yàn)室在流-固兩相流方向的的深度合作，到目前為止承擔(dān)來(lái)自美國(guó)國(guó)家科學(xué)基金會(huì)、美國(guó)能源部、國(guó)家超級(jí)計(jì)算應(yīng)用中心等政府部門(mén)以及企業(yè)的項(xiàng)目經(jīng)費(fèi)共計(jì)21,900,295美元，其中個(gè)人承擔(dān)7,613,767美元，近三年個(gè)人獲得研究經(jīng)費(fèi)共計(jì)501,212美元。Tafti 教授執(zhí)教期間共指導(dǎo)了28名博士（四名在讀）、14名博士后以及28名碩士（2名在讀），擁有豐富的研究生教學(xué)、指導(dǎo)經(jīng)驗(yàn)。
