摘要
采用案例推理技术研究了热轧带钢层流冷却数学模型中的长期自学习系数的确定方法。基于现场大量生产数据,从如何有效利用经验知识入手,对层流冷却工况和所采用的自学习系数进行案例构造,采用绝对过滤和相对过滤方法进行案例检索,根据当前工况和历史案例工况的相似度决定是否进行自学习系数的重用或修正。现场实际应用表明:对已轧过的钢种规格带钢,该方法能有效地避免再次轧制时带钢头部过冷现象,能显著提高带钢头部卷取温度的设定精度,能有效地提高换规格轧制时带钢头部卷取温度的控制精度。
Decision method for long-term coefficient used for coiling temperature control（CTC） model were studied by case-based reasoning（CBR） technology which experience and knowledge can be effectively reused based on mass production data.In the course of the study,firstly a lot of case consisted of typical laminar flow cooling conditions and self-learning coefficient adopted by CTC model were constructed,secondly case retrieval was done with the absolute and/or relative methods of filtering,then self-leaning coefficient belong to related case can be directly reused or modified according to the similarity between current conditions and historical case conditions,finally the new self-leaning coefficient was adopted into CTC model.Application show that this method can effectively avoid strip head end much more lower temperature than strip body,and can significantly improve the precision of coiling temperature control for strip head end,especially when rolling conditions or specifications changed.
出处
《钢铁》
CAS
CSCD
北大核心
2011年第12期40-43,共4页
Iron and Steel
基金
国家自然科学基金资助项目(51034009)
关键词
案例推理
自学习
层流冷却
热轧带钢
case-based reasoning
self-learning
run out table cooling
hot rolled strip
