资源动态分配项目调度问题研究与应用

发布时间：2018-04-17 23:39

本文选题：阻尼自适应粒子群算法 + 阻尼自适应惯性权重　；参考：《浙江大学》2011年硕士论文

【摘要】：项目调度问题是项目管理的重要内容,对其理论和实现方法的研究具有重要的现实意义。本文结合企业项目调度需求,提出了一种阻尼自适应粒子群算法,建立了资源动态分配项目调度问题和资源动态分配模糊项目调度问题的模型并进行了求解,开发了项目调度原型系统。最后将本文的理论和方法应用于实际注塑机开发项目的调度中,取得了良好的效果。全文的内容主要包括：第1章介绍了项目调度问题的研究背景及意义,分析了现有算法和模型中存在的问题,给出了全文主要研究内容和组织结构。第2章介绍了粒子群算法的研究概况,提出了一种阻尼自适应粒子群算法。该算法中,针对粒子群算法全局和局部搜索能力的平衡问题,对阻尼运动的模型加以改进,提出了阻尼自适应惯性权重周期性衰减的非线性改变策略；针对粒子群算法容易出现早熟收敛问题,提出了基于粒子群平均空间距离的自适应变异策略。第3章介绍了经典资源受限项目调度问题的概况,针对传统任务资源固定分配难以实现动态与高效调度,提出了资源动态分配策略：允许任务在资源未全部就绪时可以启动,任务调度期间可随资源的使用情况动态调整。给出了资源阀值的定义和当量工期的计算方法,建立了资源动态分配项目调度问题的数学模型,分析了该模型缩短工期的条件。将资源动态分配策略引入模糊项目调度问题中,建立了基于模糊工期的资源动态分配项目调度问题数学模型,模糊工期采用六点模糊数表示。分别对串行调度产生方案和并行调度产生方案进行了改进,以适应引入资源动态分配策略的项目调度问题。第4章本章将阻尼自适应粒子群算法用于资源动态分配项目调度模型和资源动态分配模糊项目调度模型的求解。提出了一种带有定界概率和定界规则的任务链表的粒子编码方法,采用基于优先规则和随机数的混合策略生成初始种群,提出了不变位交叉法对粒子实施更新、变异位领域对粒子实施变异,保证了粒子更新、变异后的可行性。对通用测试库和典型实例进行了测试,比较了不同粒子编码方法、不同资源水平和不同算法的求解效果,结果表明资源动态分配策略和阻尼自适应粒子群算法能够有效的利用资源,缩短项目工期。第5章开发了项目调度原型系统,给出了该系统的体系结构及功能模块,并将该系统应用于一个具体型号注塑机的调度过程,分析了各个模型的调度结果,得出本文提出的资源动态分配的调度策略和算法改进策略能够充分利用资源,有效的缩短项目工期。该系统在企业得到成功运行和应用。第6章总结本课题的主要研究内容和成果,展望了今后的研究方向。
[Abstract]:Project scheduling is an important part of project management.In this paper, a damped adaptive particle swarm optimization algorithm is proposed to meet the requirements of enterprise project scheduling. The models of resource dynamic allocation project scheduling problem and resource dynamic allocation fuzzy project scheduling problem are established and solved.A prototype project scheduling system is developed.Finally, the theory and method of this paper are applied to the scheduling of practical injection molding machine development project, and good results are obtained.The main contents of this paper are:Chapter 1 introduces the research background and significance of the project scheduling problem, analyzes the existing problems in the algorithms and models, and gives the main research content and organization structure.Chapter 2 introduces the research situation of particle swarm optimization and proposes a damping adaptive particle swarm optimization algorithm.In order to balance the global and local search ability of PSO, the model of damping motion is improved, and the nonlinear change strategy of damping adaptive inertial weight periodic attenuation is proposed.Aiming at the problem of premature convergence in particle swarm optimization (PSO), an adaptive mutation strategy based on the average space distance of PSO is proposed.Chapter 3 introduces the general situation of the classical resource-constrained project scheduling problem. In view of the difficulty of dynamic and efficient scheduling in traditional task resource allocation, a dynamic resource allocation strategy is proposed, which allows tasks to start when all resources are not ready.Task scheduling can be dynamically adjusted as resources are used.The definition of resource threshold and the calculation method of equivalent duration are given. The mathematical model of resource dynamic allocation project scheduling problem is established, and the conditions for shortening the time limit are analyzed.The dynamic resource allocation strategy is introduced into the fuzzy project scheduling problem, and the mathematical model of the resource dynamic allocation project scheduling problem based on the fuzzy duration is established. The fuzzy duration is represented by six points fuzzy number.The serial scheduling generation scheme and the parallel scheduling generation scheme are improved to adapt to the project scheduling problem with dynamic resource allocation strategy.In chapter 4, the damped adaptive particle swarm optimization algorithm is used to solve the scheduling model of resource dynamic allocation project and the fuzzy project scheduling model of resource dynamic allocation.In this paper, a particle coding method for task linked list with bound probability and bound rule is proposed. The hybrid strategy based on priority rule and random number is used to generate the initial population, and the invariant crossover method is proposed to update the particle.The mutation site domain implements the mutation to the particle, guarantees the particle renewal, after the mutation feasibility.The common test library and typical examples are tested, and the results of different particle coding methods, different resource levels and different algorithms are compared.The results show that the dynamic resource allocation strategy and the damping adaptive particle swarm optimization algorithm can effectively utilize the resources and shorten the project duration.In chapter 5, the prototype system of project scheduling is developed, and the architecture and function module of the system are given. The system is applied to the scheduling process of a specific injection molding machine, and the scheduling results of each model are analyzed.It is concluded that the scheduling strategy and algorithm improvement strategy proposed in this paper can make full use of resources and effectively shorten the project duration.The system has been successfully run and applied in enterprises.Chapter 6 summarizes the main research contents and results of this subject, and looks forward to the future research direction.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2011
【分类号】：TH186

【参考文献】