• CN:11-2187/TH
  • ISSN:0577-6686

机械工程学报 ›› 2026, Vol. 62 ›› Issue (18): 465-490.doi: 10.3901/JME.260624

• 交叉与前沿 • 上一篇    

扫码分享

模型降阶:数字孪生/元宇宙背景下装备性能建模的关键技术

宋学官1,2, 何西旺3, 杨亮亮4, 庞勇1,2, 阚子云1,2   

  1. 1. 大连理工大学机械工程学院 大连 116024;
    2. 高性能精密制造全国重点实验室 大连 116024;
    3. 大连民族大学机电工程学院 大连 116650;
    4. 东北电力大学机械工程学院 吉林 132012
  • 收稿日期:2025-10-30 修回日期:2026-06-11 发布日期:2026-11-20
  • 作者简介:宋学官,男,1982年出生,博士,教授,博士研究生导师。主要研究方向为装备性能建模与优化设计、装备数字孪生、整机及基础件开发(工业阀门、电铲等)研究。E-mail:sxg@dlut.edu.cn;何西旺(通信作者),男,1996年出生,博士,助理研究员。主要研究方向为结构动力学模型降阶方法、重大装备数字孪生建模关键技术。E-mail:wsxw1014@dlut.edu.cn;杨亮亮,男,1993年出生,博士,讲师。主要研究方向为面向数字孪生的信号在线处理方法、基于传感器最优布局的结构损伤识别 技术。E-mail:liangzai5358@163.com;庞勇,男,1996年出生,博士,助理研究员。主要研究方向为代理模型辅助的优化设计。E-mail:pangyong@dlut.edu.cn;阚子云,男,1991年出生,博士,副教授,博士研究生导师。主要研究方向为多体系统动力学建模、求解、软件研发、结构非线性有限元分析、机械系统数字孪生。E-mail:kanziyun@dlut.edu.cn
  • 基金资助:
    国家自然科学基金青年科学基金(52505245、52425509)和辽宁省博士科研启动基金(2025-BS-0014)资助项目。

Model Reduction: A Key Technology for Performance Modeling of Equipment in the Context of Digital Twins/Metaverse

SONG Xueguan1,2, HE Xiwang3, YANG Liangliang4, PANG Yong1,2, KAN Ziyun1,2   

  1. 1. School of Mechanical Engineering, Dalian University of Technology, Dalian 116024;
    2. State Key Laboratory of High-Performance Precision Manufacturing, Dalian 116024;
    3. School of Mechanical and Electrical Engineering, Dalian Minzu University, Dalian 116650;
    4. School of Mechanical Engineering, Northeast Electric Power University, Jilin 132012
  • Received:2025-10-30 Revised:2026-06-11 Published:2026-11-20

摘要: 随着科学研究进入第四范式,特别是步入数字孪生与元宇宙融合发展的新时代,以性能建模为核心,实现性能优化与保障、满足装备使用需求,已成为设计与制造领域的根本目标。对数字孪生与工业元宇宙装备物理状态与动态性能表征的需求,对性能建模的保真度与性能计算的实时性提出了更高要求。模型降阶作为一种通过降低复杂模型的阶数或求解复杂度,从而显著提升模型求解效率的近似方法,正逐渐成为构建高实效数字孪生与工业元宇宙的关键支撑技术与加速器。针对数字孪生与工业元宇宙对性能解算的实际需求,系统梳理了物理驱动、数据驱动及机理-数据融合驱动三类模型降阶方法的研究现状、最新进展及其典型应用,并分析了不同方法的优势与局限。在此基础上,结合两个典型应用案例,阐述了先进模型降阶方法在性能建模中的关键作用与实际价值。最后,对模型降阶方法在数字孪生与工业元宇宙中的未来发展趋势与面临的挑战进行了展望与探讨。

关键词: 数字孪生, 元宇宙, 模型降阶, 性能建模

Abstract: As scientific research enters the fourth paradigm, particularly in the era of digital twins and the industrial metaverse, equipment design and manufacturing increasingly aim to optimize and guarantee performance while meeting service requirements, with performance modelling serving as the core foundation. Digital twins and industrial metaverses, which enable the characterization of physical states and dynamic performance of equipment, impose stringent requirements on the fidelity of performance models and the efficiency of real-time performance computation. Model reduction, as an approximate methodology that decreases the order of complex models or simplifies solution processes to significantly enhance computational efficiency, has become an indispensable enabling technology for digital twins and the industrial metaverse. From the perspective of performance-solving requirements in digital twin and industrial metaverse models, the current research status, recent advances, and application practices of model reduction methods are systematically reviewed across three dimensions: physics-based reduction, data-driven reduction, and hybrid reduction. The respective advantages, limitations, and applicability of different approaches are critically analysed. Furthermore, the key role and practical value of hybrid order reduction methods in performance modelling are demonstrated through representative application scenarios. Finally, emerging trends and pressing challenges associated with model reduction for digital twins and the industrial metaverse are discussed and highlighted.

Key words: digital twins, metaverse, model reduction, performance modelling

中图分类号: