Parametric-insensitive nonparallel support vector regression for structural stress prediction of GFRP elastic gridshell structures

The gridshell structure is a type of freeform structure that is formed by the deformation of a flat grid and the final structure is a double curvature surface. The in-plane shear property and double-curvature shape create the stiffness and strength of the structure. This article aims to present a structural analysis method through a fast process by machine learning (ML) model. For gaining this purpose, design parameters including the height, width, length, and grid size of the structure are taken into consideration and the member-stresses is considered as an output. In order to obtain the stress, parametric-insensitive nonparallel support vector regression (PIN-SVR) model is considered. In this method, rather than using time-consuming finite element (FE) analysis, the PIN-SVR algorithm is applied based on generated data of FE analysis to predict the results of the structural analysis. The results show that the presented approach is an efficient method for elastic gridshell analysis

Keywords:

PIN-SVR , gridshell structure , machine learning , structural analysis , finite element