钢框架-支撑结构施工变形分析及预测方法研究OA

In this paper,the influences of welding deformation,installation error,and loading schemes on the construction deformation of braced steel frames are taken into account.A construction deformation database for such structures under different processes is established by employing the finite element method.Furthermore,four influential factors,namely welding sequence,cooling temperature,vertical deviation of columns,and loading schemes,are selected to establish a construction deformation prediction model based on the PSO-BP neural network.The accuracy of the established model is verified by using actual engineering data.The results indicate that the vertical deformation during the construction of braced steel frames is mainly controlled by welding sequence and loading schemes,accounting for 61%and 37%respectively.The lateral deformation is mainly controlled by the vertical deviation error of columns,accounting for 94%of the total lateral deformation.Compared with the actual engineering data,the established PSO-BP neural network model yields a prediction error of less than 5%,indicating that it can accurately predict the construction deformation of the braced steel frame system.This study can offer a reference for early decision-making in construction.