应用人工神经网络理论建立热粘塑性材料的本构关系

本文将人工神经网络（ＡＮＮ）用于建立热粘塑性材料的本构关系，意在探索出一种描述材料变形力学行为的新方法。文中给出了应用人工神经网络建立热粘塑性材料本构关系的ＢＰ模型和学习算法过程，并应用于４５号钢在高温和高速变形条件下的流动应力计算。计算结果与实测结果比较表明，二者吻合良好。因此，应用人工神经网络建立热粘塑性材料的本构关系具有重要的工程应用价值。

ACQUIRING THE CONSTITUTIVE RELATION FOR A THERMAL VISCOPLASTIC MATERIAL WITH ARTIFICIAL NEURAL NETWORK

n conventional constitutive theories, a kind of mathematical model is formulated to represent the plastic behavior of a material. Once the model is set 9 the material behaviors can only be expressed approximately by adjusting the parameters in the model.Artificial neural network may simulate biological nervous system and it is refered to as parallel distributed processing. It has been proved mathematically that a three--layer network can map any function to any required accuracy. So a neural network can directly map the behaviors for the thermal viscoplastic material. By the neural network, it is unnecessary to postulate any mathematical model and identify its parameters.In this paper, a four--layer backpropagation neural network is biult to acquir the constitutive relation of 45 #. Temperature, effective strain, effective strain rate are used as the input vector of the neural network 1 the output in the neural network is flow stress. After the network is trained with experimental data, it can correctly reproduce the flow stress in the sampled data. Furthermore,when the network is presented with non -sampled data, it is also can predict well. The resultes acquiring from the neural network are very encouraging.