随机工期下反应性多模式项目调度优化

Reactive Strategies in the Multi-Mode Project Scheduling with Stochastic Activity Durations

在活动工期随机中断下,通过衡量成本、鲁棒性及完工时间3类目标和两种资源分配的不同组合下,获得的反应性策略对项目调度过程中损失成本、完工时间及中断次数等方面的影响,以确定不同情况下的最优调度策略。首先对该问题进行界定,并构建问题的优化模型;设计遗传禁忌混合搜索启发式算法对模型进行求解并得出结论:遗传禁忌混合算法要优于单纯的遗传或禁忌算法;采用基于成本目标的策略可以显著降低项目反应性调度成本;在不确定度较低的环境下,传统的进度生成方式要优于随机进度生成方式。

A combination of three objectives, namely cost, robustness and project makespan, and the two kinds of resource allocation are investigated to generate six different reactive strategies. The objective is to obtain the best suitable strategy via the evaluation of adjustment cost, project makespan and the number of interruptions in the project execution process. The problem is identified first and the optimization model is constructed accordingly. Due to NP-hardness of the problem, a hybrid genetic-tabu search heuristic algorithm is developed. The proposed hybrid search heuristic outperforms the purely genetic and tabu search, and the reactive strategy based on cost minimum brings the best performance in project cost and other indexes. In low uncertain environment, the traditional schedule generation scheme is better than random schedule generation scheme.