We found a match
Your institution may have rights to this item. Sign in to continue.
- Title
Developing a visual prediction program for residual stress in girth butt welds using GA-RBF neural network.
- Authors
Jiao, Haihan; Jin, Hui
- Abstract
Residual stresses are inevitably generated during the welding process of steel structures, which can have a negative impact on the structure's normal use and fatigue performance. With the continuous advancements in intelligent manufacturing technology and modular construction of steel structures, there is a growing need for accurate prediction of residual stresses. This paper presents a visualization procedure using Python language and ABAQUS commercial software to predict the full-field residual stress of butt welds in cast steel nodes. The procedure takes into account the geometry and welding parameters of the girth butt weld as input data and provides the post-weld residual stress field as output. Instead of using the traditional thermodynamic coupling finite element calculation method, this procedure develops a parameter function form that uniformly describes the residual stress field. A radial basis function network, trained by a genetic algorithm, is employed to establish a nonlinear relationship between the input and output data. The dataset is generated through finite element simulation, and a visualization interface is created within the ABAQUS software. Experimental results validate the speed and accuracy of the proposed method. This procedure can serve as a valuable reference for quickly determining residual stress fields and optimizing welding parameters during the construction process of steel structures.
- Publication
International Journal of Advanced Manufacturing Technology, 2024, Vol 131, Issue 3/4, p1615
- ISSN
0268-3768
- Publication type
Article
- DOI
10.1007/s00170-024-13147-0