数据驱动计算力学研究进展

以数字孪生、人工智能为核心的大数据理念正深刻影响着第四次工业革4 命，数据驱动计算力学在此背景下应运而生并展现勃勃生机。与此同时，航5 空航天等尖端工业领域对高性能材料与结构的先进制造与安全评估提出了更6 严峻的挑战，经典计算力学已很难实现成倍缩短产品研发周期、实时跟踪产7 品信息并提供解决方案的目标。因此，发展面向高性能材料与结构的数据驱8 动计算力学正当其时且刻不容缓。本文拟通过梳理数据驱动计算力学的部分9 研究现状，探讨并浅析数据驱动计算力学的发展趋势.

Data-driven Computational Mechanics: a Review

The concept of big data with digital twin and artificial intelligence is profoundly influencing the fourth industrial revolution. At the same time, facing severe challenges in advanced manufacturing and safety assessment of high-performance materials and structures, classical computational mechanics has more and more limitations in achieving the goal of ‘half-time and half-cost’ of R&D cycle. Data-driven computational mechanics emerged in this context and showed great vitality. This paper aims to discuss and analyze the trend of data-driven computational mechanics by reviewing recent research achievements. In this paper, the algorithms in the framework of data-driven computational mechanics are summarized into two categories: the first one is based on energy functional, the key point of which is to construct the constitutive relationship by using material data; while the second category is based on distance functional, the specificity of which lies in directly embedding the material data into mechanical simulations. Several related data-driven algorithms of each category are briefly recalled, and the challenges and prospects of data-driven computational mechanics are discussed.