基于神经网络的金属应力状态系数模型  

Model of Stress State Coefficient in Metal Based on Artificial Neural Network

作　　者：徐如松[1] 孟令启[1] 

机构地区：[1]郑州大学机械工程学院,河南郑州450001

出　　处：《钢铁研究学报》2009年第5期55-58,共4页Journal of Iron and Steel Research

基　　金：国家自然科学基金资助项目(10176010)

摘　　要：以4200 mm轧机轧制71块钢板的实测数据为基础,利用Matlab神经网络工具箱,分别建立了轧制变形区的应力状态系数与轧前厚度、轧后厚度及轧辊直径对应关系的Elman神经网络预测模型和RBF神经网络预测模型。结果表明,所建立的两种网络模型均建立了金属应力状态系数输入和输出关系,RBF神经网络模型比Elman网络模型数据稳定,性能更优,实现了与实测结果的高度拟合。并得出不同轧辊直径对神经网络模型精度的影响规律,对轧制工艺规程的制定提出了合理建议。According to the experimental data obtained from 71 steel plates rolled in 4200 rolling mill, Elman and RBF neural network prediction models are established for the relationship between stress state coefficient and thickness before rolling, and the relationship between the thickness after rolling and diameter of roller based on Matlab neural network toolbox. The results indicate that the relationship between input and output of stress state coefficient is correctly built by the two neural networks. RBF model＇s performance is better than that of Elman model and the predicted data is highly close to the actual data. Influencing rules of model＇s accuracy are obtained when diameter of roller differs. The reasonable advice on drawing up the process specifications is proposed.

关 键 词：应力状态影响系数 神经网络 模型 

分 类 号：TG331[金属学及工艺—金属压力加工]