PREDICTION OF JOMINY HARDNESS PROFILES OF STEELS USING ARTIFICIAL NEURAL NETWORKS

Jominy hardness profiles of steels were predicted from chemical compos ition and austenitizing temperature using an artificial neural network . The neural network was trained using some 4000 examples, covering a wide range of steel compositions. The performance of the neural networ k is examined as a function of the network architecture, the number of alloying elements, and the number of data sets used for training. A w ell-trained network predicts the Jominy hardness profile with an avera ge error of about 2 HRC. Special attention was devoted to the effect o f boron on hardenability. A network trained using data only from boron steels produced results similar to those of a network trained using a ll data available. The accuracy of the predictions of the model is com pared with that of an analytical model for hardenability and with that of a partial least-squares model using the same set of data.