The flow stress of 0.4C-1.9Cr-1.5Mn-1.0Ni-0.2Mo die steel for plastic mould
s during hot deformation is predicted using the conventional regression met
hod and the artificial neural network method. The temperatures at which the
steel is compressed are 800-1100 degreesC with strain rates of 0.001-10 s(
-1) and to strains of 0-0.7. Comparisons with the result of physical tests
show that the efficiency and accuracy of the flow stress predicted using a
multi-layer perceptron network with the back propagation learning algorithm
are better than that predicted using the Zener-Holloman parameter and a hy
perbolic sine function. lt is also shown that the flow stress of die steel
for plastic moulds deformed under hot deformation conditions can be predict
ed very well using the multi-layer perceptron network with the structure of
3-9-10-1. (C) 2001 Elsevier Science B.V. All rights reserved.