基于神经网络的热连轧精轧机组轧制力高精度预报  

HIGHPRECISION PREDICTION OF ROLLING LOAD OFFINISHING STANDS WITH NEURAL NETWORKS

作　　者：吕程[1] 王国栋[1] 刘相华 姜正义 朱洪涛 袁建光[2] 解旗[2] 

机构地区：[1]东北大学 [2]宝山钢铁(集团)公司

出　　处：《钢铁》1998年第3期33-35,20,共4页Iron and Steel

摘　　要：以实测数据为基础，在精轧预设定中采用ＢＰ神经网络的方法取代传统的轧制力数学模型，并对神经网络输入项和训练样本进行分析，提出改善神经网络预报精度的一些方法。预报结果和实测数据比较表明，预报精度有较大的提高。为满足现场要求提出神经网络预报轧制力在线应用方案。On the basis of the measured data of the 2050 mm hot strip mill,the neural networks have been used for prediction of rolling load instead of traditional models for presetting finishing standsThe input of networks and training data have been analyzed and some methods for improving the precision of prediction are proposedThe comparison of the measured values and the predicted values show that the methods are very goodIn order to satisfy the demand of steel works,a proposal of application of predicting rolling load is given 

关 键 词：神经网络 轧制力预报 热连轧 轧制 精轧机组 

分 类 号：TG334.9[金属学及工艺—金属压力加工] TG331