Estimation of welding distortion using neural network

Abstract

Recently, the prediction of welding distortion of steel structure has been carried out widely. Though the thermal elastic-plastic FEM analysis is usually required to ensure the accuracy of the estimation, technical knowledge is required, and long computing time is necessary. In this study, simplification of the prediction was done using a neural network model, in which the calculation result of the FEM analysis was incorporated in teacher data, for the fillet welding of T- type build-up structure etc. This program would be able to estimate the welding distortion of the joint with different dimensions and welding conditions in a simple manner. Copyright © 2008 by The International Society of Offshore and Polar Engineers (ISOPE).