Neural networks are useful tools for optimizing material properties, considering the material's microstructure and therefore the thermal treatments it has undergone. In this research an artificial neural network (ANN) with a Bayesian framework able to predict the bake hardening and the mechanical properties of the Transformation-Induced-Plasticity (TRIP) steels was designed. The forecast ability of the ANN model is achieved taking into account the operating parameters involved in the Intercritical Annealing (IA), in the Isothermal Bainite Treatment (IBT) and also considering the different prestrain values and the volume fraction of the retained austenite before the Bake Hardening (BH) treatment. This approach allowed one to overcome the need to know the metallurgical rules that describe all the active phenomena in multiphase steels. The neural network approach allowed one to overcome the lack of prediction capability in the existing numerical models
Barcellona, A., Palmeri, D., Riccobono, R. (2009). ANN Model to predict the bake hardenability of Transformation-Induced Plasticity steels. In Materials Characterisation IV (pp. 33-45). WITpress [10.2495/MC090041].
ANN Model to predict the bake hardenability of Transformation-Induced Plasticity steels
BARCELLONA, Antonio;PALMERI, Dina;RICCOBONO, Roberto
2009-01-01
Abstract
Neural networks are useful tools for optimizing material properties, considering the material's microstructure and therefore the thermal treatments it has undergone. In this research an artificial neural network (ANN) with a Bayesian framework able to predict the bake hardening and the mechanical properties of the Transformation-Induced-Plasticity (TRIP) steels was designed. The forecast ability of the ANN model is achieved taking into account the operating parameters involved in the Intercritical Annealing (IA), in the Isothermal Bainite Treatment (IBT) and also considering the different prestrain values and the volume fraction of the retained austenite before the Bake Hardening (BH) treatment. This approach allowed one to overcome the need to know the metallurgical rules that describe all the active phenomena in multiphase steels. The neural network approach allowed one to overcome the lack of prediction capability in the existing numerical modelsFile | Dimensione | Formato | |
---|---|---|---|
34898.pdf
Solo gestori archvio
Dimensione
379.92 kB
Formato
Adobe PDF
|
379.92 kB | Adobe PDF | Visualizza/Apri Richiedi una copia |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.