Artificial Neural Network Model of ConstitutiveRelationship for 2A70 Aluminum Alloy

摘要

The hot deformation behavior of 2A70 aluminum alloy was investigated by means of isothermalcompression tests performed on a Gleeble-1500 thermal simulator over a wide range of temperatures 360~480°Cwith strain rates of 0.01~1s-1 and the largest deformation of 60％, and the true stress of the material was obtainedunder the above-mentioned conditions. The experimental results shows that 2A70 aluminum alloy is a kind ofaluminum alloy with the property of dynamic recovery; its flow stress declines with the increase of temperature,while its flow stress increases with the increase of strain rates. On the basis of experiments, the constitutiverelationship of the 2A70 aluminum alloy was constructed using a BP artificial neural network. Comparison of thepredicted values with the experimental data shows that the relative error of the trained model is less than ±3％for the sampled data while it is less than ±6％ for the non-sampled data. It is evident that the model constructedby BP ANN can accurately predict the flow stress of the 2A70 alloy.