本論文基於航機維護工廠業者之立場，考量實務營運目標與相關的限制條件，建構二個年度停機排程調整最佳化模式。本論文分成兩個研究主題進行探討。在第一主題我們考量實務上潛在隨機擾動之衝擊，設計緩衝方法，以確定性規劃建構年度停機維護長期排程調整模式。此模式可定式為一零壹整數規劃，且配合CPLEX數學規劃軟體求解。另外，本研究並發展一模擬評估方法以適當地評估所發展之確定性規劃模式與傳統確定性規劃代表性模式的停機排程在實際營運的績效。在第二主題我們參考第一主題研究架構，考量實務中航機的隨機進廠及維護時間，建構一個在實務之隨機進廠及維護時間下航機年度維護排程的調整模式。為有效地的求解實務大型的問題，本研究結合CPLEX數學規劃軟體發展一啟發解法。另外，本研究並發展一模擬評估方法以適當地評估隨機性與確定性規劃的停機排程在實際營運的績效。最後，為展示及評估此隨機規劃模式的實用績效，本論文以一國內航空公司所屬之航機維護工廠的實際維護資料為例，進行範例測試與分析。測試結果良好，顯示研究主題1之確定性模式優於傳統確定性代表模式，且研究主題2之隨機模式為三模式中最佳。其亦証實本論文二模式均為穩健且有效的規劃工具，以輔助航機維護工廠業者在面臨實際營運中的隨機航機進廠及維護時間上，每年有效地重新調整航機進廠維護排程。 In this dissertation, on the basis of the carrier’s perspective, we try to develop two optimization models for annual schedule adjustments, in accordance with the objective and related constraints required in actual operations. This dissertation is divided into two essays. In the first we study annual long-term schedule adjustments model for aircraft maintenance by deterministic planning with the prevention to deal with the impact from latent stochastic disturbances in actual operations. The model is formulated as a zero-one integer program which is solved using the mathematical programming solver, CPLEX. In addition, a simulation-based evaluation method is also developed for evaluating the schedules obtained from deterministic model of the first essay and the representative traditional deterministic model in simulated real operations. In the second we consider framework of the first essay to develop an optimization model for annual schedule adjustments under stochastic aircraft check-in and maintenance times in actual operations. To effectively solve the model with realistically large problems occurring in real world, a heuristic is developed with the use of the mathematical programming solver, CPLEX. In addition, a simulation-based evaluation method is also developed for evaluating the schedules obtained from the stochastic and deterministic models in simulated real operations. Finally, to demonstrate and to evaluate the two models in practice, we individually perform a case study using the real operating data from a major aircraft maintenance center in Taiwan. The test results are good, showing that deterministic model of the first essay is better than the traditional deterministic model and the stochastic model of the second essay is the best of the three models. It is also showing that the two models would be robust and useful planning tools for aircraft maintenance centers to effectively adjust their aircraft maintenance schedules annually under stochastic aircraft check-in and maintenance times in actual operations.
