Prediction of the flow stress of high-speed steel during hot deformation using a BP artificial neural network

Abstract

The hot deformation behavior of T1 (W18Cr4V) high-speed steel was investigated by means of continuous compression tests performed on a Gleeble 1500 Thermomechanical simulator over a wide range of temperatures (950–1150°C) with strain rates of 0.001–10 s −1 and true strains of 0–0.7. The flow stress under the above-mentioned hot deformation conditions is predicted using a BP artificial neural network. The architecture of the network includes three input parameters: strain rate ε ̇ , temperature T and true strain ε; and just one output parameter: the flow stress σ. Two hidden layers are adopted, the first hidden layer including nine neurons and the second 10 neurons. It has been verified that a BP artificial neural network with 3–9–10–1 architecture can predict the flow stress of high-speed steel during hot deformation very well. Compared with the prediction method of flow stress using the Zener–Holloman parameter and hyperbolic sine stress function, the prediction method using the BP artificial neural network has higher efficiency and accuracy.