PREDICTION OF YIELD STRENGTH OF LOW/MEDIUM CRMO FERRITIC STEELS USING ARTIFICIAL NEURAL NETWORKS

Abstract

AbstractThe yield strength of Low/Medium Cr- Mo ferritic steels has been analyzed by a well selectedartificial neural networks (ANN) model using data sets obtained from ASTM publications. Thequalitative and quantitative effects of chemical composition, heat treatment and test temperaturehave been studied. The proposed ANN model was obtained by applying averaging process to thefirst best three models. The first one consists of 24 input nodes (the input variables), 23 hiddennodes and the output node which is the target for the required yield strength. Among the previousvariables, it was found that the heat treatment ones have the greatest contribution to the yieldstrength especially the tempering one i.e. the average contribution of about 15% was obtained