【作者】 褚春超；

【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】 针对目前项目进度管理中普遍存在的进度超期、预算超支、关键资源在需要时不能满足、资源冲突严重等问题,论文应用概率理论、模糊理论和约束理论,借助网络计划技术、相依结构矩阵和计算机仿真方法,对工程项目进度管理中计划制定、工序工期不确定性、资源制约下进度规划、工序复杂关系处理和有效缩短项目工期等问题分别进行了系统的研究。在项目计划制定方面,本文对项目计划制定中常存在的计划片面、流于形式、过粗或过细、没考虑风险、缺乏裕量等问题进行了剖析,强调了项目相关人员直接参与制定计划的必要性和重要性,并提出了制定项目计划的具体实施步骤。针对工序工期的不确定性,将其看作是各不确定因素影响下的结果,建立了基于影响因素的进度模型,不但可以灵活描述不确定的工序工期,而且能揭示工序工期间在不确定因素作用下的相关依赖性,突破了传统网络计划方法中把工序工期看成是相互独立变量的作法。另外,针对工序工期的模糊不确定性和传统模糊网络计划方法在求取工序时差时出现的问题,提出了改进的模糊网络进度模型。改进的模型将概率统计方法引入到模糊网络计算中来,提出了模糊数量化和模糊数大小比较的原则,并给出了模糊数运算的新定义;在完成工序最早时间参数计算的基础上,直接计算工序的时差,避免了出现负值时间参数。针对现有资源制约规划(Resource-Constrained Scheduling, RCS)方法下工序时差不准确、关键路径不连续、工序执行顺序不稳定、规划方案单一以及对工序间资源链关系识别不全面等问题,提出了改进的基于资源链的RL-RCS(Resource Link-based RCS)模型。它通过对资源在工序间使用、转移情况的跟踪来建立工序间的资源链,保证了对工序间资源依赖关系识别的完整性,并就所建立资源链的必要性和其对工序时差计算的影响进行了分析。进一步,还在识别替代规划方案和多余工序关系方面做了相应探讨。为了解决现有网络计划方法描述工序关系单一,且不能揭示工序执行中的返工循环属性及学习性等问题,提出了基于相依结构矩阵(Dependency Structure Matrix, DSM)的项目规划模型。在模型中,采用工序关系矩阵R-DSM(Relationship of DSM)不但可以描述工序间的顺序和并行关系,而且还可以灵活地描述工序间的搭接依赖和时间窗口式依赖关系;考虑到工序执行过程中的返工循环特性和返工时的学习性,提出了采用概率矩阵P-DSM(Probability of DSM)来描述工序执行过程的合理性;给出了应用平行离散事件模拟仿真方法计算项目工期的具体实施步骤。针对关键链项目管理(Critical Chain Project Management, CCPM),提出了确定工期裕量的新方法。该方法综合考虑了项目资源紧张度、网络图结构复杂度和管理者风险偏好等因素的影响,能确保确定的工期裕量适当。进一步,对实施关键链项目管理给出了具体步骤,并对其中的工序工期估计、工期裕量估计及设置与监控、关键链的确定与更新、并行任务和管理观念转变等问题做了深入分析。最后,本文借助VisualStudio.NET语言开发了相应的软件,给出了面向对象技术的系统结构设计和类模型框架,并对系统相关的单代号网络图生成技术,以及与Access数据库、AutoCAD软件和Excel表格的接口技术进行了介绍。更多还原

【Abstract】 Nowadays there are some problems in project scheduling management, such as time or cost overflowing, critical resource scarcity or conflicting, and so on. So some theories and methods including probability theory, fuzzy theory, theory of constraint (TOC), network planning technique, dependency structure matrix (DSM), and simulation technique are inducted to treat with those problems in the dissertation. By virtue of these, some aspects on drawing scheduling, duration indeterminacy, resource-constrained scheduling (RCS), complicated activity relationships and shorting project period are separately comprehensively studied.About drawing scheduling, some frequent mistakes including incomplete, superficial, more cursory or more specific, lacking in risk appraisal, inadequate for buffer sizing are deeply analyzed in the dissertation. The fact is emphasized that people related to project should be required participating directly, and concrete steps of drawing scheduling are presented.About indeterminacy of activity durations, effect factors-based stochastic network planning model is proposed, which looks upon indeterminate activity durations as every independent factors effecting results. This method not only may describe activity durations stochastic indeterminacy but also reveal dependent relationships among activity durations under the same indeterminate factors effecting, and breaks out hypothesis on activity durations independency in traditional models. Otherwise, improved fuzzy network planning model is proposed to cope with fuzzy activity durations and shortcoming of computing floats in traditional methods. Statistical mean is introduced to quantify and compare different fuzzy number, and put forward new definitions of fuzzy number operations in the improved method. Having calculated activity’s earliest time parameters, activity’s floats are worked out directly, which avoids embarrassment of negative time parameters.About resource-constrained scheduling (RCS), an improved resource link-based resource-constrained scheduling (RL-RCS) model is proposed. By pursuing resource used in and transferred among activities resource links among activities are set up in the model, with integrity of recognizing dependency on resources guaranteed, and inaccurate activity floats, inconsecutive critical path, unstable activity executing sequence, singleness of scheduling, and incomplete recognition of activity resource links all solved. Additional analysis of necessity of setting up resources links and influence of resource links on calculating activity floats is made. Furthermore, investigations are carried out into recognition of substitutions and redundant relationships among activities. Because of some drawbacks of current network planning techniques in disability to describe rich relationships among activities and to reflect potential iteration and learning attribute during activity execution, dependency structure matrix (DSM) -based project scheduling model is proposed. In the model, activity relationships of DSM (R-DSM) is employed to describe not only sequential and parallel relationships but also overlapping and time-window constraint-dependent relationships among activities, and activity probability of DSM (P-DSM) is used to compatibly describe iteration and learning ability during activity execution. At last, concrete procedures are given to calculate project period by virtue of parallel discrete events-based simulation method.About critical chain project management (CCPM), an improved buffer sizing method is proposed. Effects of project resource intensity, network structure complexity and managers’attitude to risks on buffers are all analyzed and integrated in the method, which makes buffers sizing even more applicable. Furthermore, concrete steps are given to execute CCPM, and comprehensive analysis of CCPM is made, with activity duration estimated, buffer estimated, inserted and monitored, critical chain established and updated, multi-tasking implemented and management concepts changed.At last, related software is developed in VisualStrudio.NET language, and system frame and class structure model based on object-oriented ideas are given. Moreover, some techniques are also demonstrated including layout of activity-on-node network by computer aided design, interface programs with Access database, AutoCAD software and Excel table.更多还原