Data-driven mechanical modeling based on the first principles of elasticity
Received:August 11, 2023  Revised:October 08, 2023
View Full Text  View/Add Comment  Download reader
DOI:10.7511/jslx20230811001
KeyWord:structural mechanics model  data-driven method  equation identification  Timoshenko beam  first principles
                 
AuthorInstitution
郑勇刚 大连理工大学 工程力学系 工业装备结构分析优化与CAE软件全国重点实验室 大连
吴哲同 大连理工大学 工程力学系 工业装备结构分析优化与CAE软件全国重点实验室 大连
张涵博 大连理工大学 工程力学系 工业装备结构分析优化与CAE软件全国重点实验室 大连
刘振海 大连理工大学 工程力学系 工业装备结构分析优化与CAE软件全国重点实验室 大连
叶宏飞 大连理工大学 工程力学系 工业装备结构分析优化与CAE软件全国重点实验室 大连
张洪武 大连理工大学 工程力学系 工业装备结构分析优化与CAE软件全国重点实验室 大连
Hits: 400
Download times: 294
Abstract:
      A data-driven mechanical modeling method based on the first principles of elasticity is proposed in this paper, which enables the identification of concise mechanical models that can accurately capture the deformation mechanism from the numerical simulation results of elasticity.Based on the high-precision data obtained from finite element calculations and the unsupervised data-driven identification method of Seq-SVF, the governing differential equation to describe the bending deformation in the form of Timoshenko beam is automatically identified from the load and displacement data, and the expressions of the shear correction coefficient as a function of structural dimension and mechanical parameters under three different loading conditions are identified.The results reveal the applicability of classical beam models in cases of different loading conditions, and a new model is also established.By combining the first-principle calculation of elasticity with the data-driven paradigm, a new approach is provided for building and analyzing complex mechanical models, which can overcome the limitations and the strong dependence on human experience of traditional modeling methods.