Abstract: To reduce the project time and optimize the resource utilization,the multi-mode resource-constrained project scheduling problem with activity overlapping was proposed.The overlap and rework activity time factor matrix was designed to describe the relations between overlap and rework,and the multi-mode resource-constrained project scheduling optimization model with activity overlapping was proposed whose objective was to minimize the project's duration.A revised genetic algorithm based on specific schedule generation mechanism was presented to solve the model.Based on the classic double list encoding genetic algorithm,a preprocessing stage of the activity list was designed to improve its efficiency.The model and algorithm were demonstrated by using an example project and comparative experiments.The duration of project was tested and analyzed through full factorial design,which provided basis for decision makers.

Key words: project scheduling, resource-constrained, multi-mode, overlapping, genetic algorithms

摘要： 为了缩短项目工期、优化资源利用效率,研究了带有活动重叠的多模式资源受限项目调度问题,构建了活动重叠—返工时间因子矩阵,对多模式下的活动重叠和返工时间进行了完整的数学描述,以最小化项目工期为目标,建立了带有活动重叠的多模式项目调度优化模型;设计了改进的遗传算法并对问题进行求解,在经典的双链编码遗传算法的基础上,设计对初始种群活动链的预处理阶段以加速算法的求解效率,并针对多模式活动重叠问题设计了专门的解码方法。通过实验研究验证了该算法较其他方法具有更好的求解能力和表现,以一个小规模算例演示了模型在处理工期缩短问题上的有效性。通过全因子实验设计分析了问题参数对缩短项目工期的影响,为项目管理者确定项目调度方案提供了决策依据。

关键词: 项目调度, 资源受限, 多模式, 活动重叠, 遗传算法