The application of neural networks to analysis of the effects of chemical composition on hardenability of steel

Purpose: The goal of the research carried out was evaluation of alloying elements effect on the development of artificial neural network models, allowing the determination of the Jominy hardenability curve based on the chemical composition of constructional and machine steels. Design/methodology/approach: MLP neural network was used to learn rule for modelling the steels properties. Then the neural network used for computer simulation synergistic effect of alloying elements on the hardenability of steel. Research limitations/implications: Results of the research confirmed that neural networks are a useful tool in evaluation the effect of alloying elements on the properties of materials compared to conventional methods. Additionally it confirms idea, that based on data from standards and catalogues is possible to develop the assumed model. Practical implications: It has been demonstrated complete the practical usefulness of the developed models in the selection of materials designed machine parts, which allows the direct relationship during the melting process real time control of the desired hardness of the steel hardenability curve. Originality/value: Based on the results of catalogues and standards with the used of neural networks developed and fully validated experimental model of the relationship between hardenability and chemical composition of the constructional and machine steels.