基于神经网络的金属变形抗力预测

Prediction of metal flow stress by neural network

以恒变速率凸轮压缩试验机得到的实验数据为基础,利用Matlab人工神经网络工具箱,建立了优质碳素结构钢的变形抗力预测模型。通过该网络的μ参数的自适应调整,提高了收敛速度,使金属塑性变形抗力的预测精度大为提高。

Based on the data obtained from cam plastometer, this paper calculated the model of flow stress in high quality carbon steel. The calculating method uses the self-suit adjustment of the parameter μ in the network to increase the convergence speed and then to improve the prediction accuracy of metal flow stress.