【作者】 郭海滨；

【导师】 郑丕谔；

【作者基本信息】 天津大学 ， 管理科学与工程， 2007， 博士

【摘要】 工程项目的建设需要使用大量的资源,如何对这些资源进行合理的配置是工程项目管理领域的一类非常重要的问题。研究工程项目资源配置模式的目的,就是要通过研究先进的模型和方法,来实现对资源的优化配置。对此,本文建立了四种不同的模型来改进资源配置的效果。作业可间断的多资源均衡配置问题是受到广泛关注的问题。在本文建立的模型中,通过引入作业间断率和间断延期率等无量纲变量来描述作业间断后的安排情况,可以使作业间的逻辑约束关系在模型求解过程中得到自动满足。该模型还可以兼容传统模型中作业全部不可间断和特殊作业不可间断的情况。另外,还通过引入时差风险等概念来改进模型,从而避免资源均衡过程中对作业时差的过度利用。资源强度可变的多资源均衡模型,可用来求解允许作业资源强度变化的资源均衡问题。在该模型中,允许适当延长作业的工期来降低作业的资源强度,并引入工期延长率这一无量纲变量来表示作业工期所延长的程度。该变量使得模型中各作业间的逻辑关系约束在求解过程中能自动满足。该模型也可以兼容作业资源强度全部不可变化以及特殊作业资源强度不可变化的情况。另外,通过对该模型加以改进,也可以在资源均衡过程中对作业的工期风险加以考虑。基于扩展时差的多资源受限的单项目调度模型用来求解满足资源约束的单项目调度计划的最小工期。通过定义一种新的扩展时差的概念,可以囊括工期取值不同的各种调度计划方案。通过引入无量纲化的扩展工时率变量和罚函数,则可以避免求解得到不满足作业间逻辑关系或资源约束的非可行解。模型的求解可以直接得到调度计划的安排,从而避免了通过一定的调度生成方式由作业链表生成调度计划的过程。进一步,对本文的单项目调度模型加以改进后,得到用于多资源受限多项目调度问题的模型,并以最小化各项目的加权总工期为目标函数进行求解,实现了对多项目环境下有限资源的优化配置。最后,通过采用遗传算法对本文提出的各类模型进行求解并对结果加以对比分析,验证了这些模型的有效性。更多还原

【Abstract】 A large amount of multi-resources need to be utilized in the building-up process of construction project. It is an important problem in the research field of project management that how resources are allocated reasonably. The research objective of resources allocation mode is to optimize resources allocation by developing and utilizing advanced models and algorithms. To improve effects of resources allocation, four models are developed in this dissertation.A multi-resources leveling model allowing activity break is established to solve a kind of multi-resources leveling problems, in which non-critical activities can be broken for some time during leveling process. Some dimensionless variables such as activity breaking rate and breaking delay rate, are introduced to describe the instance of activity break in this model. With the help of these variables the logical relation constraints among activities can be automatically satisfied during the whole computing process. The model can also realize the same function as conventional models without activity break and can fit the instance allowing break of special activities. Moreover, the model is improved by introducing some concepts such as float time risk in order to avoid the overuse of activity float time in resources leveling process.A multi-resources leveling model with changeable resource intensity is founded to solve another kind of multi-resources leveling problem, in which prolonging of activity duration is allowed to decrease resource intensity of activity. A dimensionless variable named as duration prolonging rate is introduced to show the prolonged degree of activity duration. Logical relation constraints among activities can be automatically satisfied with the help of this variable during the whole computing process. The model can also realize the same function as conventional models with unchangeable resource intensity and can fit the instance allowing unchangeable resource intensity of some special activities. The model is also improved in order to take into account the time risk in resource leveling process.A multi-resources constrained project scheduling model based on extended float time is developed to calculate the minimized finishing time of a single construction project schedule with resources constraints satisfied. Valuing range of activity starting time can be broaden by defining a new kind of concept named extended float time. Thus, various project schedules with different complete time can be considered. The infeasible solutions violating either logical relation constraints among activities or resources constraints can be avoided through introducing a dimensionless variable named extended float time rate and utilizing penalty function. Project schedule can be directly obtained through results of this model, with no need of the process that a project schedule must be brought into being from a list of ranked avtivities.A multi-project scheduling model with multi-resources constraints is obtained by improving the scheduling model for single project mentioned above. With the objective function being changed to the minimization of weighted total duration, the model can realize optimal allocation of constrained resources for multiple projects.Models in this dissertation are all verified with the help of genetic algorithm and results comparison.更多还原