改进型复杂过程全局进化算法在热连轧负荷分配中的应用  

Load distribution optimization of hot strip mills with improved evolutionary algorithm for complex-process optimization

作　　者：胡长斌[1] 童朝南[1] 彭开香[1] 

机构地区：[1]北京科技大学钢铁流程先进控制教育部重点实验室,北京100083

出　　处：《控制与决策》2012年第1期15-21,27,共8页Control and Decision

基　　金：国家自然科学基金项目(60374032);北京市教委重点学科项目(XK100080537)

摘　　要：为实现热连轧精轧机组负荷分配的优化设定,提出一种具有柔性框架结构的改进型复杂过程全局进化算法.该算法部分地借用了分散搜索原则,在通用框架中嵌入具有搜索机制的子方法;利用无限折叠映射混沌模型和局部搜索法,分别对初始种群的生成和"超越"深度搜索进行改进以提高最优解的求解效率.实验结果表明,该算法能够使用较少的参数完成负荷分配优化问题的可行解搜索,具有较好的时效性,是局部和全局搜索的有机体.An improved evolutionary algorithm for complex-process optimization（IEACOP） is presented to achieve optimal setting of load distribution about hot rolling mill. The algorithm is partially based on the principles of the scatter search, which has flexible structure and is embedded in sub-methods that have search mechanism. Meanwhile, infinite folding chaotic model and local search method are applied to improve initial population strategy and ＂go-beyond strategy＂ of in- depth search. Then the efficiency of the local optimal solution is improved. The experiment results show that IEACOP makes use of fewer adjustable parameters to get feasible mathematical solution for the actual load distribution problems and validate the real-time application, which is the organism including local search and global search.

关 键 词：负荷分配 复杂过程优化 分散搜索 进化方法 

分 类 号：TG333.5[金属学及工艺—金属压力加工] TP29[自动化与计算机技术—检测技术与自动化装置]