【摘要】 为适应单件、小批、个性化和市场需求动态多变的特点,研究了多种不同交货期生产订单并存时的可重构制造系统中虚拟制造单元构建问题。构造了以连续加工产品间相似系数之和最大、工作加班时间最少、单元的封闭性最好(工件跨单元搬运次数最少)、制造系统重构成本最小及设备生产负荷均衡为目标的非线性多目标0-1整数规划模型。采用两阶段的求解策略进行求解。在第一阶段采用启发式方法对非瓶颈设备和工艺进行预处理,以缩小问题解的搜索空间;第二阶段采用一种基于网格计算的分布式平行协同多目标粒子群算法,随机搜索Pareto优化解集。最后,利用globus 4.0工具箱搭建计算网格和Java语言实现了算法。从生产实际出发给出了算例,证明结果可行,从而验证了算法的有效性。更多还原

【Abstract】 To adapt to characteristics as single workpiece,small-batch,diversified and dynamic market demand changes,a methodology which could be used to form Virtual Manufacturing Cell(VMC) in reconfigurable manufacturing system for multiple product orders with different due dates was proposed.A non-linear multi-objective 0-1 integer programming model was constructed.The model was aimed to maximize the sum of conjoint processing products’ similarity coefficient,minimize the working overtime,maximize the independence of manufacturing cells(minimize the inter-cell movements of parts),minimize the cost of reconfiguration of virtual manufacturing cells and minimize the machine loading unbalance.Two-phase resolution strategy was applied in this methodology to figure out the model.The primary objective of Phase I was to compress the searching space by pre-assignment of non-exceptional machine tools and non-exceptional processes with a heuristic method.To search the Pareto front in Phase II,a new Distributed Parallel Collaborative Multi-Objective Particle Swarm Optimization(DPCMOPSO) algorithm based on grid computing was employed.Finally,a computing grid was constructed with globus 4.0 toolkit,the algorithm was developed with java.The effectiveness and feasibility of algorithm were verified through an example.更多还原