Data-Based Models Predicting Residual Stress in the Welding Zone of Dissimilar Metals and Their Uncertainty Analysis

Since welding residual stress is a major factor in the generation of Primary Water Stress Corrosion Cracking (PWSCC), it is essential to examine the welding residual stress to prevent PWSCC. In order to predict this residual stress, several artificial intelligence methods have been developed and used as a powerful tool in nuclear engineering fields. In this study, two data-based models, support vector regression (SVR) and fuzzy neural network (FNN), were used to analyze the residual stress for dissimilar metal welding under a variety of welding conditions. The data was obtained in a previous study [1] by performing FEAs under various welding conditions, such as pipeline shapes, welding heat input, welding metal strength and the constraint of the pipeline end parts. This paper deals partly with regression models using FNN [1] and SVR [2] to easily predict the residual stress in the dissimilar metal welding for pipelines at nuclear power plants (NPPs). This paper also builds on previous studies [1-2] to analyze the uncertainty of a residual stress prediction using artificial intelligence methods.