Development of computational system based on artificial neural network for prediction of high temperature deformation behaviour in steels

A significant number of the data about the hot deformation behaviour of the metallic materials is published up to date in the scientific sources. As a result, the systematization and formalization of such big datasets using math-ematical modeling is required. However, the accuracy of the usual regression models is not enough for this purpose. The artificial neural network (ANN) based model for prediction of the steel flow stress under the hot deformation conditions was constructed. The model has shown a high accuracy on the training dataset such as on the independ-ent additional compression tests of the stainless 13Cr11Ni2W2MoV steel. The compression tests were carried out in the strain rate range of 0.1, 1, and 10 s-1 and the temperature range of 1000 - 1200 oC using thermomechanical simulator Gleeble 3800. The steel has a two-phase ferritic/austenitic microstructure in the hot deformation range. The average relative errors for the training dataset and approvement tests were about 8.8 and 9.5 %, respectively. The constructed model may be used for the determination of the different element influence the flow stress of the steel, which may allow fast correction of the hot deformation conditions. Special software was developed for the use of the built ANN-based model.