无模拉拔过程中金属线材直径的BP神经网络预测模型  

Prediction model of wire diameter in dieless drawing process based on BP neural network

作　　者：杨勇强[1] 何勇[1] 刘雪峰[1] 谢建新[1] 

机构地区：[1]北京科技大学材料科学与工程学院,北京100083

出　　处：《塑性工程学报》2008年第1期118-122,共5页Journal of Plasticity Engineering

基　　金：国家重点基础研究发展计划资助项目(2006CB605200);国家自然科学基金资助项目(50474072;50634010);长江学者和创新团队发展计划资助项目(IRT0407)

摘　　要：利用Matlab软件建立了无模拉拔成形过程中金属线材直径的BP神经网络预测模型,通过实验验证了网络模型的可靠性。网络预测值与实测值之间的平均相对误差为0.52%,最大相对误差为2.42%,具有较高的预测精度。将所建模型用于无模拉拔成形过程中线材尺寸的在线控制,具有明显效果,在较大程度上减小了线材直径沿长度方向的波动。Using Matlab software, the prediction model of wire diameter in dieless drawing process is established based on BP neural network method. The experimental results show that the prediction model is effective and reliable. The average relative error between the predictions and the experimental results is 0. 52%, and the maximal relative error is 2. 42%. Adopting the developed model to the on-line controlling of wire diameter in dieless drawing process, obvious control effect is obtained, and the fluctuation of wire diameter along the length is reduced to a great extent,

关 键 词：无模拉拔 神经网络 线材直径 预测模型 在线控制 

分 类 号：TG356.4[金属学及工艺—金属压力加工] TP183[自动化与计算机技术—控制理论与控制工程]