Prediction of Steel Hardenability and Related Reliability through Neural Networks

Silvia Cateni, Valentina Colla, Marco Vannucci, and Marco Vannocci


Neural Networks, Steel, Hardenability


Hardenability is an important property of steel, that is characterised through a curve known as Jominy profile. The paper presents a neural network-based approach to the prediction of this curve on the basis of steel chemical composition. The proposed approach is divided in two parts: firstly a set of neural networks is exploited for the sequential estimation of the points of the curve. Afterwards an additional network estimate the reliability of the profile prediction at each point. The model proved to be effective through tests on a real industrial dataset.

Important Links:

Go Back