变结构PI迭代学习及在层流冷却过程中的应用  

Iterative learning method based on variable-structure PI and its application in laminar cooling process

作　　者：片锦香[1,2] 柴天佑[3,4] 李界家[1] 

机构地区：[1]沈阳建筑大学信息与控制工程学院,沈阳110168 [2]中国科学院沈阳自动化研究所信息服务与智能控制研究室,沈阳110016 [3]东北大学自动化研究中心,沈阳110189 [4]流程工业综合自动化国家重点实验室,沈阳110189

出　　处：《仪器仪表学报》2013年第11期2478-2483,共6页Chinese Journal of Scientific Instrument

基　　金：国家自然科学基金(61273031);住建部科学技术计划(2012-K7-19);沈阳建筑大学青年基金(2012038)资助项目

摘　　要：热轧带钢层流冷却过程具有强非线性、工况条件变化剧烈、难以采用精确数学模型进行过程描述的复杂工业特性,这些特性决定了需要新的控制技术,以保证带钢最终质量。层流冷却过程的重复性使得迭代学习方法适用于冷却带钢之间(批次之间)的学习,因此提出变结构PI迭代学习方法对层流冷却过程进行设定控制,采用案例推理技术根据变化工况调整PI参数。实验研究结果证明,该方法可以对类似操作条件的带钢进行有效迭代学习,并很快找到工作点,使冷却结束后的带钢卷取温度控制在目标范围内。The hot rolled strip laminar cooling process has the features of strong nonlinearity and severely changed op- erating conditions, and these complex industrial features are difficult to describe with an accurate mathematic model. As a result, a new control technology is needed to ensure the final quality of the strip. The repeat characteristics of the laminar cooling process make the iterative learning method suitable for the learning between the strips （between the batches）. So, the iterative learning method based on variable-structure PI is proposed to carry out the preset and con- trol of the system in this paper,in which the parameters of PI are adjusted with the CBR （ cased-based reasoning） technology according to the varying working conditions. The experiment research results show that the proposed meth- od can find the right operating point quickly for the strips with similar working condition, and make the coiling tem- perature of the strips after the cooling process controlled within the target range.

关 键 词：层流冷却 迭代学习 变结构PI 卷取温度 案例推理 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]