摘要
将源于BP人工神经网络改进策略的动量项技术引入到带钢热连轧负荷分配中来,提出了误差补偿迭代算法,介绍了该算法的推导过程,并对算法的合理性进行了分析。基于误差补偿迭代算法及传统的Newton-Raph-son法,开发了带钢热连轧精轧过程设定系统软件。离线模拟计算结果表明,与Newton-Raphson法相比,误差补偿迭代算法对初始厚度分布的要求比较宽松,有利于新产品的开发,具有良好的在线应用前景。
The technique of momentum item which comes from improved strategy of BP（back propagation） artificial neural networks was introduced into load distribution for tandem hot strip rolling, and then a new iterative algorithm named error-comp iterative algorithm was put forward, its derivation process was introduced, the rationality of algorithm was analyzed. Based on the new algorithm and Newton-Raphson algorithm, the set-up system software for finish rolling process was developed. The off-line simulation result shows that the new algorithm places less demands on the initial thickness than the conventional Newton-Raphson algorithm, and thus is more helpful for developing new products, it is also well fit for on-line application.
出处
《钢铁》
CAS
CSCD
北大核心
2006年第2期46-50,共5页
Iron and Steel
关键词
负荷分配
计算机过程控制
动量项
带钢热连轧机
误差补偿迭代算法
load distribution
process computer control
momentum item
tandem hot strip mill
error-comp iterative algorithm
