面向流体力学的多范式融合研究展望

Translated title of the contribution: Prospects of multi-paradigm fusion methods for fluid mechanics research

Research output: Contribution to journalArticlepeer-review

16 Scopus citations

Abstract

Experimental observation, theoretical research and numerical simulation are the basic research paradigms in many disciplines, including fluid mechanics. Since the 21st century, artificial intelligence based on big data has become an important driving force, leading to new scientific and technological revolution and industrial transformation. This is known as the data-intensive scientific research paradigm, which forms the fourth research paradigm. Similarly, data-driven machine learning has also become an emerging research direction in fluid mechanics and promoted the progress in intelligent fluid mechanics. However, compared to traditional data-intensive research paradigm that relies on "Internet and big data", research on intelligent fluid mechanics has its own unique background. For example, compared to high-dimensional flow state, geometric boundary conditions, and the inherent high-dimensional, cross-scale, random, and nonlinear characteristics of complex flow, research in data-driven fluid mechanics essentially handles large data but small samples. Although there are three major research paradigms in fluid mechanics, the integration among research paradigms is very low, where engineering optimization simply corrects data from multiple sources. Multi-source data fusion can alleviate several dilemmas, like small data sample from a single source, difficulties in modelling, and the insufficient utilization of low-fidelity data, it still fails to fully integrate theoretical models, expert knowledge and experience from the basic paradigms.Therefore, based on the fourth paradigm driven by artificial intelligence, the organic combination of three major research topics including experiment, theoretical model and numerical simulation, developing date and knowledge jointly driven multi-paradigm fusion methods for fluid mechanics, have become urgent to solve major practical engineering problems, as well as to satisfy the need for the development of the connotation and the characteristics of fluid mechanics in the new era.

Translated title of the contributionProspects of multi-paradigm fusion methods for fluid mechanics research
Original languageChinese (Traditional)
Pages (from-to)433-467
Number of pages35
JournalAdvances in Mechanics
Volume53
Issue number2
DOIs
StatePublished - Jun 2023

Fingerprint

Dive into the research topics of 'Prospects of multi-paradigm fusion methods for fluid mechanics research'. Together they form a unique fingerprint.

Cite this