以多能工概念求解多技能資源限制專案排程問題之研究

資源限制專案排程問題一直以來都是各界須直接面臨之問題，隨著產業環境的不同所需面臨的排程問題有其不同的複雜性，在現今工業4.0智能製造環境中，同一專案在多個地方之間存在著不平等的狀況，尤其是在處理眾多複雜的作業時，這些作業活動受到許多關係和條件的限制，其中勞動力技能的異質性會導致產品及服務品質的變化。另外，由於大多數應用整數規劃之研究皆為人力排班相關，並未針對專案排程的領域。因此，本研究將所能運用之『人力資源』視為專案可用之『可再生資源』，且具不同專業特質和熟悉度之人員編制，針對專案作業中來自不同技能的需求，工作項目的各種資源的組合即為多組態模式，使執行時間亦有所不同，此為多能工概念下的多技能RCPSP問題。以專案完成時間最小化為目標，採整數規劃法提出一適合多能工之資源限制排程模型，同時建置以多能工為基礎的資源限制下專案排程之測試例題，該測試例題包含七種作業規模(亦即j10- j30)，並以不同的技能資源配置及組態置換兩種方法來測試該數學模型是否適合多能工之資源限制專案排程問題且分析當中會影響專案完成時間之因素。結果顯示，配置隨機技能在分配至所需作業時彈性空間較大，配置集中技能時資源調整空間較小；而在組態置換中，原始組態與替換不同組態後之專案完成時間並無太大的差異。

關鍵字

資源限制專案排程問題；整數規劃法；多能工；資源配置

並列摘要

Resource-Constrained Project Scheduling has always been an unavoidable decision-making problem in various industries, especially when dealing with many complex operations bounded by numerous and complicated resources, relationships and conditions. There has been some inequality in scheduling decisions because workforce skill heterogeneity leads to variation in end-product/service quality. Since most of the research on the application of integer programming is related to workforce scheduling, it does not focus on the area of project scheduling. Thus, we will discuss project scheduling, where activities require skills from several specializations at different levels. We regard these workers as valuable, renewable, and discrete resources with multiple different skills. They are sometimes assigned to perform task-sequence activities collaboratively and collectively to minimize the project completion time. We establish a mathematical model which enhances such competitive advantage and solve this problem through the integer programming method. Furthermore, we generate problem instances based on multi-skilled workers. These problem instances include seven activity sets, i.e. j10, j12, j14, j16, j18, j20, and j30. In the end, we consider two different skill allocations and mode replacements to see any potential impact on the project completion times. The result indicates that flexibility for configuring random skills is greater when allocated to the required tasks, while resource adjustability is lesser when configuring centralized skills. Moreover, there is no visible difference in project completion time between the original configuration and the replacement of different configurations in configuration replacement.