【作者】 黄辉；

【导师】 庞南生；

【作者基本信息】 华北电力大学 ， 工业工程， 2012， 硕士

【摘要】 广义网络调度问题实质是在广义网络时序关系下通过活动进度的合理安排使得项目关注的目标得到优化。传统的项目调度问题研究往往是在无延迟网络,仅考虑可更新资源的基础上对项目工期进行优化。实际上它忽视广义网络下的活动时序关系、活动的执行模式、资源类别等其他因素对活动进度安排的影响。另外,随着企业对项目目标要求的提高,在项目目标优化上,项目管理者并不只是一味地追求工期最短,还关注其它目标,如资源均衡程度、抗风险能力等等。为此,针对传统项目调度安排研究中有关活动执行模式和活动间的时序关系假设条件和缺陷,本文研究了活动多执行模式、广义网络关系、可更新资源和不可更新资源两种资源约束对调度安排的影响,建立了综合考虑工期、时间鲁棒性、资源均衡多目标分层优化模型。同时,对求解模型的算法也进行了进一步地研究,将串行进度生成机制与遗传算法、模拟退火算法相结合,利用优先权数作为编码,采用串行进度生成机制作为解码方式对目标函数进行求解,并对遗传退火过程中的交叉机制和多目标选择策略进行了分析,形成了新型的遗传退火算法,并通过算例对模型进行了求解分析,验证模型和算法的可行性。在基本完成预期研究目标的同时,本文也发现了研究过程中存在的问题,如何克服这些不足,通过更为科学的方法建立项目进度目标体系,以及将算法与禁忌等其它算法结合起来寻求更好地解决项目调度问题的算法都将是今后值得思考的方向。更多还原

【Abstract】 The essence of the generalized network scheduling problem is through the reasonable arrangement of the progress of the activities to make the goal of the project to optimization in generalized network timing relationships.The traditional research on project schedule problems usually thinks that the net of project schedule is a no-delay network and the activities’resources are all renewable resources. In fact, it ignored the timing relationships in the generalized network, different execution mode of the activities, resource type and so on.With the increasing demand for the project objectives, project managers not just blindly pursue the shortest duration, but also concerned about other goals, such as resource equilibrium level, the ability to resist risk.Therefore, Aiming at the defects of the assumptions of execution mode for activities and net relations between activities,the paper studies the influence of the multiple execution mode, generalized network relations, two resource constraints(the renewable resources and the non-renewable resources) on scheduling arrangements, and build multi-objective hierarchical optimization model for the time limit, time robustness and resources equilibrium. At the same time, the algorithm to solve the model is the further studied, which is combined serial schedule generation mechanisms with genetic algorithms and simulated annealing algorithm.The algorithm uses priority number as a code and Serial schedule generation mechanism as the way of decoding to solve the objective function.In addition, the crossing mechanisms and the strategy of multi-objective selection are further Analyzed in the process of genetic simulated annealing in the paper.Then, through examples for solving the model analysis, the feasibility of the proposed model and algorithm is verified.While achieving the expected objectives basically, this paper also finds some problems during the studying. How to build a more perfect target system by scientific methods and combine the annealing with the other algorithm (taboos algorithm) to seek better way to solve project scheduling problem, will be a researching direction worth considering.更多还原