在傳統經濟生產批量(economic production quantity; EPQ)的模式中,假設生產系統是完美的,但生產設備隨著時間的增加進而導致機台的可靠度降低進而造成當機,本研究考慮經濟生產批量模型,而機台為不可靠模組的製造系統並且決定生產批量。該機台考慮二個不可靠關鍵零組件,本研究分為兩大主要模式探討(一)故障發生率為隨機發生(二)加入變動生產率作為探討。在故障發生率為隨機發生的狀況下,此二個關鍵零組件有三種故障來源,其失效平均時間皆為指數分配。另一模式為依附生產率所衍生的變動生產率狀況下,依然是有二個關鍵零組件且有三種故障來源,其失效平均時間亦皆為指數分配。本研究採用Groenevelt et al.所研究之生產模型考慮維修策略及生產運作之門檻值(threshold value)之外,另亦考慮2種生產策略:No-resumption (NR)及Abort-resume (AR)。在數值分析中探討不同參數對近似解之影響,另以基因演算法結果作比較,最後以目前產業界可以應用的範圍作為探討,並且以機械手臂作為探討對象。
This study considers an Economic Production Quantity (EPQ) model that is applied to determine a production lot size for manufacturing a product in batches on an imperfect machine to meet a constant demand rate. The machine is dictated by two unreliable key components that may be random breakdown during a production run. There are two situations to handle the defects in this study.(1)subject to random failure(2)subject to variable production rates. In the model for random failure that suppose there are three independent sources of shocks. It occurs at a random time all following an exponential distribution. In the model for variable production rates that also suppose there are three independent sources of shocks. It occurs at a random time all following an exponential distribution. There two production policies, the no-resumption (NR) and the abort-resume (AR), which differ from Groenevelt et al. research work by considering the repair activity and a threshold value of the production uptime, are concerned in this paper. Various properties and near-optimal formulas are further discussed. And numerical examples are provided to argue these approximations’ quality in comparison with results obtained by the genetic algorithm. The practical application of a real world example can be seen in an automatic robotic arm machine, in which a hydraulic cylinder and a barrel heater are two key components.