不确定知识化制造环境下航空发动机装配车间滚动自进化  

Rolling Self-evolution of an Aero-engine Assembly Shop in Uncertain Knowledgeable Manufacturing Environment

作　　者：姜天华[1,2] 严洪森[1,2] 汪峥[1,2] 

机构地区：[1]东南大学自动化学院,南京210096 [2]东南大学复杂工程系统测量与控制教育部重点实验室,南京210096

出　　处：《机械工程学报》2017年第1期165-173,共9页Journal of Mechanical Engineering

基　　金：国家自然科学基金(60934008);中央高校基本科研业务费专项资金(2242014K10031);江苏高校优势学科建设工程资助项目

摘　　要：针对产品装配次数不确定且装配组的调整时间与工序间排序相关的航空发动机装配车间,对不确定环境下知识化制造系统(Knowledgeable manufacturing system,KMS)的自进化问题进行研究。采用事件和周期混合驱动型的自进化机制,结合滚动时域方法实现航空发动机装配车间自进化,并给出一种可行的滚动规则。建立系统在各个决策时刻的静态决策子问题的数学模型,并针对该模型提出一种具有双层结构的遗传算法进行求解。在下层的混合型遗传算法中,给出一种直接解码算法,并引入了变邻域搜索算法,以增强局部搜索的能力。通过仿真对算法的性能进行了测试,此外,试验数据表明执行自进化操作的系统具有较好的生产性能。尤其是对于更敏感于自身调整的系统,自进化操作发挥的作用更大。For an aero-engine assembly shop with uncertain number of assemblies and sequence-dependent setup times of work groups, the self-evolution problem of knowledgeable manufacturing systems （KMS） in uncertain manufacturing environment is studied. Rolling horizon method is combined with the hybrid event-driven and periodic-driven self-evolution mechanism to implement the self-evolution of the aero-engine assembly shop, and a feasible rolling rule is given. A mathematical model of the static decision sub-problem at each decision moment is established. For the model, a two-level genetic algorithm is proposed to solve it. In the lower-level hybrid GA, a direct decoding algorithm is given, and the variable neighborhood search algorithm is introduced to enhance the ability of local search. The performance of the algorithm is tested by simulations. In addition, numerical experiments show that the system with self-evolution operations has a better production performance. Self-evolution plays a much larger role especially for the system which is particularly sensitive to the self-adjusting.

关 键 词：不确定知识化制造环境 自进化 航空发动机装配车间 滚动时域 遗传算法 变邻域搜索 

分 类 号：TH165[机械工程—机械制造及自动化]