基于人工神经网络的中碳含钒微合金钢热变形流变应力预报
Prediction on Hot Deformation Flow Stress of Medium-carbon Micro-alloyed Steels with Vanadium Based on Artificial Neural Network
吴晋彬1, 刘国权1,2, 许 磊1, 王承阳1
点击:3595次 下载:0次
作者单位:(1. 北京科技大学 材料科学与工程学院, 北京 100083; 2. 北京科技大学 新金属材料国家重点实验室, 北京 100083)
中文关键字:钒微合金钢;人工神经网络; 流变应力; 预报
英文关键字:vanadium-micro-alloyed steel; artificial neural network; flow stress; prediction
中文摘要:以0.33C,0.40Si,1.50Mn,0.099V(wt%)的中碳含钒微合金钢在应变速率为0.005~30 s-1、温度为750~1050 ℃条件下的单向热压缩变形实验数据为样本数据,用商用软件matlab6.5构建BP人工神经网络模型。经实验数据验证,该模型预测的流变应力结果可靠。研究结果表明:利用人工神经网络方法建立热变形流变应力预测模型,适用于预测一定温度与应变速率范围内(0.1~0.9)应变处的热变形流变应力,为控制轧制工艺参数提供参考。与常用的表征稳态或峰值应变处的流变应力与温度和应变速率关系的Arrhenius方程相比,应用范围更广。
英文摘要:The models of BP artificial neural network were established using commercial software matlab6.5 on medium-carbon micro-alloyed steel with 0.33 wt% carbon, 0.40 wt% silicon, 1.50 wt% manganese, 0.099 wt% vanadium, which were performed on Gleeble-1500 thermal simulator during uniaxial hot compression deformation tests at 750~1050 ℃ and strain rate of 0.005~30 s-1. The flow stress predicted by the established models is conisitent with the experiment data. The results show that the established artificial neural network models are applicable to characterize flow stress in the strain rate of 0.1~0.9 during hot deformation, it can provide a reference for the control of rolling process parameters. The model can be used in a wider range of the strain compared with arrhenius equation which is often suitable to static or peak strain.