資源撫平(Resource Leveling)問題自從資源限制專案排程(Resource Constrained Project Scheduling Problem,RCPSP)發展以來就一直倍受重視，有許多的相關研究在解決這方面的問題。可是現有的資源撫平問題，都著眼在單一模式資源限制專案排程(Single-mode Resource Constrained Project Scheduling Problem,SRCPSP)上，在現實事件的專案中，單一模式的專案非常少見。而另依方面，再現有的多模式資源限制專案排程問題中，亦未見到相關資源撫平的文獻出現，因此本篇論文希望能將資源撫平的問題拓展於多模式資源限制專案排程問題(Multi-mode Resource Constrained Project Scheduling Problem,MRCPSP)之上。 本研究以資源撫平為目標，利用遺傳演算法同時進行作業模式(Mode)的選擇以及排程的結果。此外，利用遺傳演算中適應函數(fitness function)的調整，使得此演算法可以依照管理者不同需求而去調整適應函數以達到預期中希望達到的目標。 The fluctuation of required resources causes problem cause problem in project scheduling. Thus research developed resource leveling techniques to minimize the deviation between the resource requirement and the desired resourced profile. However, the resource leveling problem, can not be considered with Multi-mode Resource-Constrained Problem (MRCPSP), that is an activity can be preformed in one or more combinations of durations and resource requirements. In this paper, we consider resource leveling in Multi-mode Project Scheduling. We describe a Genetic Algorithm for this problem. The proposed algorithm can effectibly provide the near-optimal solution for different situation and fit the goal that is project manager wanted.
