摘要
针对中厚板在轧后层流冷却过程中采用传统自学习方法修正计算对流换热系数存在的不足,提出利用BP神经网络对自适应系统进行改进。预报结果表明,采用神经网络计算换热系数后,终冷温度的控制偏差在±15℃以内,明显提高了终冷温度的控制精度,具有在线应用的前景。
To deal with the deficiency of the self-learn method of heat transfer coefficient during plate laminar cooling process,a BP neural network was put forward to improve the self adapting system. The results indica ted that the difference between the calculated temperature and target finish cooling temperature was controlled between -15℃ and +15℃ after heat transfer coefficient was predicted by BP neural network. The method improves the accuracy of finish cooling temperature obviously, so it has an anticipant application on line in the future.
出处
《轧钢》
2007年第2期45-47,61,共4页
Steel Rolling
基金
国家自然科学基金(50634030)
关键词
中厚板
换热系数
自学习
人工神经网络
plate
heat transfer coefficient
self-learn
BP neural network