面向薄壁复合梁结构的数据驱动计算均匀化方法研究OACHSSCDCSTPCD

A research on the data-driven computational homogenization method for thin-walled composite beam structures was conducted,and an efficient data-driven multi-scale model based on model reduction technology was developed.Firstly,in terms of macro-micro scale correlation calculations,the data-driven computational homogenization method decouples the traditionally nested macro-micro scale calculations.It is replaced with a micro-offline and macro-online scheme,where an offline database of stress-strain equivalence for representative volume elements drives the online computational efficiency.Secondly,for the macroscopic model,a reduced-order model of thin-walled composite beam structure is constructed based on Carrera's unified formula,which significantly reduces the degrees of freedom and the number of integration points.Finally,in terms of data searching,a hierarchical data searching scheme is adopted to replace the exhaustive data searching method to reduce the data volume per search.The study shows that the data-driven computational homogenization method combined with model reduction techniques can significantly improve the computational efficiency while ensuring accuracy,providing a reliable and efficient tool for the rapid prediction of multi-scale mechanical behaviors of thin-walled composite beam structures.