基于LM算法延时反馈BP网络的铝热连轧板凸度预报  

The predictive model of aluminum strip's crown at hot tandem rolling mill's finisher delivery based on BP network with LM algorithm

作　　者：龚寄[1] 骆拓[2] GONG Ji LUO Tuo(Mechanical Engineering College, Hunan International Economics University, Changsha 410205, China Changsha Research Institute of Mining and Metallurgy, Changsha 410012, China)

机构地区：[1]湖南涉外经济学院机械工程学院,湖南长沙410205 [2]长沙矿冶研究院,湖南长沙410012

出　　处：《轻合金加工技术》2016年第11期19-22,共4页Light Alloy Fabrication Technology

摘　　要：铝热连轧机是一个极其复杂的系统,为了精确地建模预报末机架出口处的板凸度,提出了利用延时反馈BP网络对板凸度时间序列的预报方法,来实现对铝带板凸度的精确建模和预测。通过某"1+4"铝热连轧机现场板凸度的预报试验,采集了大量试验数据。结果表明,板材相对凸度的预报精度可达到0.01%。验证了该方法的有效性和优越性。Aluminum hot tandem rolling mill is an extremely complicated system. In order to predict the crown of aluminum strip at the finisher delivery more accurately, this article has put forward a method using the time-lapse feedback BP network to forecast the time-se- quencing of the crowns, and the reconfiguration of BP network was trained with LM algo- rithm. A large number of test data were gathered from the on-the-spot experiment of a ＂1 ＋ 4＂ aluminium hot tandem rolling mill, to simulate and verify this method. The simulation re- sults show that the predictive precision of relative crown（ % ） of the strip can reach 0.01%, indicating an available and advanced method.

关 键 词：铝热连轧机 板凸度 延时反馈BP网络 LM算法 时间序列预报 

分 类 号：TG333[金属学及工艺—金属压力加工]