基于多品种小批量订货型的平行机分批调度方法研究

发布时间：2018-03-29 13:41

本文选题：分批　切入点：调度　出处：《广东工业大学》2011年硕士论文

【摘要】：多品种小批量订货式生产在制造业中占据着很重要的地位。柔性制造单元(FMC)是用于多品种、中小批量生产的具有高柔性且自动化程度高的制造系统。近年来,许多中心批量生产的企业已经将此单元应用在实际生产中。交货期是订货型企业的生命线。在车间安排生产任务时,多难以兼顾多订单的交货期要求。FMC的生产环境下,为降低单元的使用成本,提高设备的利用率,可以将不同订单的相同类型工件组成若干批次进行加工。组批和调度方案的好坏直接影响着企业经济效益和声誉。因此,对这样的实际问题如何编制合理优化的组批调度方案是非常有必要的。本文以一家大型轮胎模具生产企业为背景,研究一类集成批量计划和平行机调度的问题,该问题具有订单交货期、到达时间和加工准备时间等约束。首先在订单到达时间确定的情况下,建立单个数学模型描述集成问题,以降低单元加工费用和订单拖期惩罚费用为目标；提出一种带启发式规则的遗传模拟退火两阶段算法(GASA)。算法引入启发式规则生成基础批有效减少了染色体长度,从而加快搜索速度。遗传算法对基础批进行全局搜索,在批量确定的情况下,模拟退火进行局部搜索,得到当前分批情况下的优值。然后以此模具企业的实际生产例检验该算法的收敛性,证明该算法在可以接受的时间内是有效可行的；通过对比GASA和GA算法的求解效果,说明相比于GA算法,这种带启发式规则的混合算法确实能够更快更好的求得问题的较优解；选取五种不同规模的实际生产例进行数值仿真,分别采用该算法和一种经典算法在相同的计算时间内进行求解。对计算结果对比分析表明随着是任务规模的增大,该种算法的优势更加明显,从而说明了该模型和算法针对这一类特殊问题更为有效和可行。接着下一步,针对上述分批调度数学模型,考虑到实际生产中的订单到达时间不确定因素,且订单的到达过程符合泊松过程,从而建立期望值模型,在随机变量的概率密度函数已知的情况下,将问题模型按照确定性模型来处理,采用前文所提带启发式规则的遗传模拟退火两阶段算法进行求解。通过大规模的数值仿真实验,证明该算法对待订单到达时间不确定问题的求解同样有效,并且这一类问题的求解过对订程单到达时间不敏感。最后,运用先进的建模和仿真工具eM-Plant,结合应用面向对象的方法,进一步深入面向平行机的分批调度模型。利用eM-Plant中的SimTalk语言对各个对象及活动进行编程控制,还原现实生产环境。对比仿真模型运行结果和GASA结算结果,证明GASA在现实生产中具有可行性。
[Abstract]:The flexible manufacturing unit (FMC) is a manufacturing system with high flexibility and high automation, which is used in many varieties, medium and small batch production. Many enterprises in the center of mass production have applied this unit to actual production. Delivery time is the lifeline of an order-oriented enterprise. In order to reduce the cost of unit use and improve the utilization rate of equipment, it is difficult to take account of the production environment of multi-order delivery time requirement .FMC when the workshop arranges production tasks. The same type of workpieces of different orders can be assembled into several batches for processing. The quality of the group batch and scheduling scheme directly affects the economic efficiency and reputation of the enterprise. It is necessary to work out a reasonably optimized batch scheduling scheme for such practical problems. Based on the background of a large tire mould manufacturing enterprise, this paper studies a class of integrated batch planning and parallel machine scheduling problems, which have order delivery time. First of all, a single mathematical model is established to describe the integration problem in order to reduce the unit processing cost and the penalty cost of the order delay when the order arrival time is determined. A two-stage genetic simulated annealing algorithm with heuristic rules is proposed. The heuristic rule is introduced to generate the base batch, which can effectively reduce the chromosome length and speed up the search. In the case of batch determination, simulated annealing performs local search to get the best value in the current batch condition. Then the convergence of the algorithm is verified by the actual production examples of the die and mould enterprises. It is proved that the algorithm is effective and feasible in the acceptable time, and by comparing the results of GASA and GA algorithms, it is proved that the hybrid algorithm with heuristic rules can obtain the optimal solution of the problem faster and better than GA algorithm. Five practical production examples of different scales are selected for numerical simulation. The algorithm and a classical algorithm are used to solve the problem in the same time. The comparison and analysis of the calculated results show that with the increase of the scale of the task, The advantages of this algorithm are more obvious, which shows that the model and algorithm are more effective and feasible for this kind of special problems. Considering the uncertainty of order arrival time in actual production, and the arrival process of order accords with Poisson process, the expected value model is established. When the probability density function of random variable is known, the probability density function of random variable is known. The problem model is treated by deterministic model, and the genetic simulated annealing two-stage algorithm with heuristic rule is used to solve the problem. It is proved that the algorithm is equally effective in solving the uncertain order arrival time problem, and that the solution of this kind of problem is not sensitive to the arrival time of single order. Finally, the advanced modeling and simulation tool eM-Plantis used in combination with the object-oriented method. Further deeper into the batch scheduling model for parallel machines. Using the SimTalk language in eM-Plant to control each object and activities, restore the real production environment, compare the results of the simulation model and GASA settlement, compare the results of the simulation model and the results of GASA settlement, and compare the results of the simulation model with the results of GASA settlement. It is proved that GASA is feasible in practical production.
【学位授予单位】：广东工业大学
【学位级别】：硕士
【学位授予年份】：2011
【分类号】：TH186

【引证文献】