Open AccessRESEARCH ON THE PREDICTION OF DEFORMATION PROPERTIES BASED ON NEURAL NETWORK
Author(s) -
Lu Jin,
Chunli Zhang,
Xiangyuan Meng
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/2078/1/012057
Subject(s) - artificial neural network , structural engineering , deformation (meteorology) , rigidity (electromagnetism) , frame (networking) , finite element method , displacement (psychology) , column (typography) , buckling , beam (structure) , computer science , materials science , engineering , artificial intelligence , mechanical engineering , composite material , psychology , psychotherapist
The direct analysis method based on deformation properties needs to predict the lateral deformation of the steel frame-support structure in advance, but it takes time and labor to use both finite element numerical simulation and experimental research. In this paper, the BP neural network receives the parameters of the steel frame-support structure, such as the aspect ratio, the second-order effect parameters, the slenderness ratio of frame columns, the rigidity ratio of beam-column bus and the deformation limit, with the lateral displacement of the column as the output layer, the second-order lateral deformation properties of the steel frame-support structure can be quickly obtained. Comparing the prediction results of BP neural network with those of the existing fitting formulas, the absolute difference and the relative error ratio between them are very small, which shows that the accuracy of the second-order column roof migration using the BP neural network prediction is high. The BP neural network can be used to judge the applicability of the direct analysis method based on the deformation properties of the steel frame-support structure under various parameters.