Applications of Multi-Objective Genetic Algorithms for Real-World Optimization in Production Scheduling

[[abstract]]在本文中，針對製造業目前所面臨的全球性競爭以及非預期性需求波動問題，試圖改善其生產排程的方式。因為多數製造業的生產排程主要是由現場管理者（Shop-Flow）來管控，但因其技術與經驗有限，所規劃的生產排程表並不合適。 另一方面，現場管理者必需同時考量多項的排程績效目標，且所關切的排程績效目標常具有衝突性及損益交換之特性（Trade-Off）。也就是說，欲提升某一目標之績效則會致使另一目標之績效下滑。 在本文中，針對於二個案例公司中的三個實務案例，利用多目標遺傳基因演算法進行生產排程最佳化，並且成功協助現場管理者有效率的求解真實製造環境的生產排程表，提供適時且適合的決策。[[abstract]]Global competitions and problems of the irregular demand pattern are challenging the manufacturing industry nowadays and utilization of multi-objective genetic algorithm in production scheduling is suggested to improve the efficiency of traditional method in this research Most of the manufacturing production scheduling is mainly controlled by shop-flow and scheduling results are not always proper or fully satisfied due to lack of professional techniques and experiences On the other hand shop-flow is also required to take several scheduling goals of efficiency into considerations and the goals are often contradictory That is to say that improving one of the goals would result in decreasing another goal of efficiency In this research two case companies and three practical examples are represented to demonstrate how applications of multi-objective genetic algorithm in optimizing production scheduling successfully helped shop-flow to find the most efficient scheduling result and meanwhile provide the most proper solutions for decision maker