Motivated by the recent trend in delivering projects with value or benefit to stakeholders and seeking to reduce the significant fraction of projects plagued by schedule and budget overruns, researchers are looking at lean project management (LPM) as a possible solution. This paper outlines a new approach to project scheduling in an LPM framework. We develop and solve a math program for balancing project time, cost, value, and risk, seeking to maximize the project value subject to schedule and budget constraints in multimode stochastic projects. Each activity mode contains fixed and resource cost information and duration data, and may be associated with one or more value attributes, thereby integrating project and product scope. By selecting a mode for each activity, the value of the project is determined, and stability is achieved by complying with on-schedule and on-budget probability thresholds. We solve the problem by applying a reinforcement learning-based heuristic, a tool known for obtaining fast solutions in a variety of applications in uncertain environments. We validate the method by comparing the results to two benchmarks—those obtained by solving a mixed-integer program, and the values obtained by adapting a recently published genetic algorithm. Our method generates competitive values faster than the benchmarks, making this approach interesting for the planning stage of a project, when multiple project tradespace alternatives are explored and solved, and runtime is limited. Our approach can be applied by decision-makers to calculate an efficient frontier with the best project plans for given on-schedule and on-budget probabilities.

受最近向利益相关者交付具有价值或利益的项目的趋势的推动，并寻求减少因进度和预算超支而困扰的大部分项目，研究人员正在将精益项目管理 (LPM) 视为一种可能的解决方案。本文概述了在 LPM 框架中进行项目调度的新方法。我们开发并解决了一个数学程序，用于平衡项目时间、成本、价值和风险，在多模式随机项目中寻求在进度和预算约束下最大化项目价值。每个活动模式都包含固定和资源成本信息和持续时间数据，并可能与一个或多个价值属性相关联，从而整合项目和产品范围。通过为每个活动选择一种模式，确定项目的价值，稳定性是通过遵守按计划和按预算的概率阈值来实现的。我们通过应用基于强化学习的启发式方法来解决这个问题，这是一种以在不确定环境中的各种应用中获得快速解决方案而闻名的工具。我们通过将结果与两个基准进行比较来验证该方法——那些是通过求解混合整数程序获得的，以及通过调整最近发布的遗传算法获得的值。我们的方法比基准更快地产生有竞争力的价值，使这种方法在项目的规划阶段很有趣，当探索和解决多个项目贸易空间替代方案时，运行时间是有限的。决策者可以应用我们的方法来计算有效边界，并针对给定的按计划和按预算概率提供最佳项目计划。