物料限制下面向订单的动态生产调度问题研究

发布时间：2018-07-14 20:41

【摘要】：面对全球制造业向智能制造的方向发展的情况下,依据客户的需求导向安排生产的趋势在当今市场更为明显,客户的需求越来越多样化和个性化,为了满足客户要求,企业逐步从传统面向库存的生产方式转变为面向订单的生产方式。在这样的背景下,本文的研究问题考虑了物料限制的约束条件,将物料供给和生产环节协同调度,又因为订单的准时交付是重要服务标准,所以本文研究问题是以最小化总拖期为目标函数的物料限制下面向订单的动态生产调度问题。核心制造商要求接收所有随时到达的订单,根据订单向仓库发出物料需求信号,随后根据物料的库存情况向上游的供应商发出订货通知,最后根据物料的供给状况制定生产计划。因为订单的随机性,生产调度方案要求随着新订单的到达发生相应变化,所以本文研究问题是典型的动态调度问题。当动态事件(接受新订单)发生时,将未完成加工的产品和新订单产品重新调度,根据当前加工状态重新产生一个调度方案。首先,分析研究问题,提出问题的合理假设,依据问题本身的限制提出约束条件,并解释数学约束的含义,以总拖期最小化为目标函数,建立了混合整数规划模型。然后,提出两种改进的万用启发式算法,分别为改进人工免疫系统算法(Modified Artificial Immune System algorithm,MAIS)和多变邻域搜索算法(Multiple Variable Neighborhood Search,MVNS)。在理解人工免疫系统算法的设计思想和基本原理的基础上,汲取了自然免疫系统记忆处理信息的优点改进免疫系统算法。改进人工免疫系统算法的结构由V(D)J基因重组、体细胞超突变、类型转换和二次免疫反应构成。多变邻域搜索算法由初始化过程、四种邻域结构、扰乱过程、邻域变换以及强化过程构成,其中邻域变换过程使其可以自动修正邻域结构以跳出局部最优解的空间,强化结构加快算法的收敛速度,扰乱机制使其从局部最优解连续没有改善的情况挣脱以寻得更好的可行解。最后,计算机仿真实验分析和评价各种算法解决该问题的性能。随机组合问题参数产生240个实验问题,采用C++语言编程算法,通过相对百分比偏差、运算时间、标准差以及假设检验对算法的性能做出比较,验证改进人工免疫系统算法和多变邻域搜索算法能够在短时间内解决研究问题得到近似最优解。
[Abstract]:In the face of the global manufacturing industry developing towards the direction of intelligent manufacturing, the trend of arranging production according to the customer's demand is more and more obvious in today's market, the customer's demand is more and more diversified and individualized, in order to meet the customer's demand, Enterprises gradually change from the traditional inventory-oriented production mode to the order-oriented production mode. In this context, the research of this paper takes into account the constraints of material constraints, the supply of materials and production links coordinated scheduling, and because the timely delivery of orders is an important service standard. Therefore, this paper studies the dynamic production scheduling problem with the objective function of minimizing the total tardiness. The core manufacturer requires that all incoming orders be received, that a material requirement signal be sent to the warehouse according to the order, and then an order notification be sent to the upstream supplier based on the stock of the material. Finally, according to the supply of materials to formulate production plans. Because of the randomness of orders, the production scheduling scheme needs to change with the arrival of new orders, so the problem is a typical dynamic scheduling problem. When the dynamic event (accepting new order) occurs, the unfinished product and the new order product are rescheduled, and a scheduling scheme is re-generated according to the current processing state. Firstly, by analyzing and studying the problem, putting forward the reasonable hypothesis of the problem, putting forward the constraint condition according to the limitation of the problem itself, and explaining the meaning of the mathematical constraint, the mixed integer programming model is established with the objective function of the total delay minimization as the objective function. Then, two improved universal heuristic algorithms, modified Artificial immune system algorithm (mais) and multiple variable neighborhood search algorithm (MVNS), are proposed. On the basis of understanding the design idea and basic principle of artificial immune system (AIS) algorithm, the advantages of natural immune system (NIS) memory processing information are derived, and the immune system algorithm is improved. The structure of the improved artificial immune system is composed of V (D) J gene recombination, somatic hypermutation, type conversion and secondary immune response. The algorithm consists of initialization process, four kinds of neighborhood structure, disturbance process, neighborhood transformation and reinforcement process, in which neighborhood transformation process can automatically modify neighborhood structure to jump out of the space of local optimal solution. The reinforcement structure accelerates the convergence speed of the algorithm, and the perturbation mechanism makes it break free from the continuous condition that the local optimal solution is not improved to find a better feasible solution. Finally, the performance of various algorithms to solve the problem is analyzed and evaluated by computer simulation experiments. Random combinatorial problem parameters generate 240 experimental problems. C language programming algorithm is used to compare the performance of the algorithm by relative percentage deviation, operation time, standard deviation and hypothesis test. It is verified that the improved artificial immune system algorithm and the variable neighborhood search algorithm can solve the problem in a short time and obtain the approximate optimal solution.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TB497

【相似文献】