應用人工智慧於多機多途程最適生產排程之研究

AN OPTIMAL APPLICATION OF ARTIFICIAL INTELLIGENCE ON PRODUCTION SCHEDULING UNDER MULTI-MACHINE AND MULTI-ROUTING

本研究主要探討多機零工型工廠之生產排程問題，在多機多途程之順序不一之假設下，考慮各機台加工之間具有非對稱之整備時間，使用隨機優化演算法（Random Optimization Algorithm，簡稱ROA），將演算法導入人工智慧領域中之智慧代理人（Intelligent Agents，簡稱IA）之理論，以四種績效指標為目標，即總完工時間最小化、總延遲天數最小化、總延遲罰鍰最小化與總延遲訂單數最小化等，可依使用者之需求選擇單一目標進行求解。此外，並建置一智慧型排程系統，加入實務上之需求，新增VIP客戶以及臨時插單等功能，以利提升整個生產效率。將數據結果與常見之派工法則（FIFO、EDD）當作初解去做比較，研究結果證實本研究提出之演算法能有效改善排程績效。 This paper focused on the job shop scheduling problem of multi-machine and multi-routing when sequences are different. In consideration of the asymmetrical set-up time with each machine, this paper used a Random Optimization Algorithm (ROA) and Intelligent Agents (IA) that is in the area of Artificial Intelligence (AI). The multi-objective function minimizes the completion time (makespan), delay times, penalty, and tardiness. The users can choose one goal to solve this problem. In addition, this paper also constructed an intelligent scheduling system and added some features such as VIP customers and inserting temporary orders. The results were compared to dispatching rule (FIFO and EDD) as an initial solution. The effectiveness of the proposed approach is demonstrated through an experimental evaluation.