面向混合流水线的任务智能调度系统的研究与实现

发布时间：2018-03-05 04:24

本文选题：服装生产　切入点：任务智能调度　出处：《东华大学》2012年硕士论文　论文类型：学位论文

【摘要】：作为一类劳动密集型的制造业,服装生产的任务调度与控制主要依赖于生产管理者的知识、经验甚至直觉。这导致生产决策往往不是最优并且是不可靠的。同时,随着小批量生产的流行与产品款式的频繁变化,单靠人工进行生产任务调度已无法满足当前服装快速生产的要求。本文针对服装生产中的任务调度要求,对生产任务调度问题进行了分析与建模,提出了基于一类改进的离散差分进化算法的任务智能调度模型。在此基础上,开发了集离散事件仿真模型与任务智能调度模型于一体的生产调度专家系统,为服装敏捷制造过程中的生产计划与控制提供科学的决策。本论文研究的主要贡献和创新点如下 1.根据服装混合流水线的特点对任务调度问题进行数学建模,给出了其目标函数与约束条件。针对多目标的总体评价问题,提出了效用函数用于综合多种效用值,并给出了效用函数的定义。 2.提出了利用离散事件仿真模型实现对混合流水线动态过程的模拟,解决了无数学方程可直接获取混合流水线动态性能目标的难题。此外离散事件仿真模型还能为管理者提供直观的方式用于微调调度策略,为管理者手动分配任务提供一种便捷的途径。 3.提出一种基于Pareto最优集思想的改进离散差分进化算法,该算法有效地增加了种群的多样性,解决了传统差分算法过早收敛的问题。同时,将改进的离散差分算法用于服装生产中的任务调度问题。大量实验结果表明了该算法的正确性和有效性。 4.开发了一类适用于服装企业手工制衣流水线的任务智能调度专家系统,给出了其程序结构及主要模块的实现过程,并对其应用结果进行分析与讨论。本文提出的基于Pareto选择策略的离散差分进化算法,以及基于此算法的任务智能调度专家系统,将使国内服装企业受益于由科学的任务分配策略所带来的生产敏捷性和生产效率的提升,增强它们在全球服装行业中的竞争优势。
[Abstract]:As a kind of labor-intensive manufacturing industry, task scheduling and control of garment production mainly depend on the knowledge, experience and even intuition of the production manager. With the popularity of small batch production and the frequent changes of product styles, manual production task scheduling can no longer meet the requirements of rapid garment production. Based on the analysis and modeling of production task scheduling problem, a task intelligent scheduling model based on a class of improved discrete differential evolution algorithm is proposed. A production scheduling expert system which integrates discrete event simulation model and task intelligent scheduling model is developed to provide scientific decision making for production planning and control in garment agile manufacturing process. The main contributions and innovations of this paper are as follows. 1. According to the characteristics of mixed pipeline, the task scheduling problem is modeled, and its objective function and constraint conditions are given. In view of the multi-objective overall evaluation problem, the utility function is used to synthesize various utility values. The definition of utility function is given. 2. A discrete event simulation model is proposed to simulate the dynamic process of mixed pipeline. In addition, the discrete event simulation model can provide a direct way for managers to fine-tune the scheduling strategy, and solve the problem that the mixed pipeline dynamic performance target can be obtained directly without mathematical equations, and the discrete event simulation model can provide an intuitive way for managers to fine-tune scheduling strategy. It provides a convenient way for managers to assign tasks manually. 3. An improved discrete differential evolutionary algorithm based on the idea of Pareto optimal set is proposed. The algorithm effectively increases the diversity of population and solves the problem of premature convergence of traditional difference algorithm. The improved discrete difference algorithm is applied to the task scheduling problem in garment production. A large number of experimental results show that the algorithm is correct and effective. 4. A kind of task intelligent dispatching expert system suitable for garment manufacturer pipeline is developed. The program structure and the realization process of the main modules are given, and the application results are analyzed and discussed. In this paper, a discrete differential evolutionary algorithm based on Pareto selection strategy and a task intelligent scheduling expert system based on this algorithm are proposed. It will enable domestic garment enterprises to benefit from the improvement of production agility and production efficiency brought about by scientific task allocation strategy and enhance their competitive advantage in the global garment industry.
【学位授予单位】：东华大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TH186

【参考文献】