| 摘要: | 良好的飛航排程規劃，除需考量航空公司本身的機隊供給、相關作業措施、競爭中市場佔有率外，亦需考量市場需求在營運中的不確定性。以往短期飛航排程之研究多以草擬班表為輸入，並大多基於預測之平均需求值，進行機隊排程與班次表建立，忽略了實務營運上，每日的旅客需求可能的隨機變動情形。近來雖有研究針對市場競爭建立短期變動需求飛航排程模式，然其仍假設每日的市場需求固定。此假設忽略了現實每日營運的隨機需求現象，而此等隨機性需求可能降低原規劃的最佳化排程與班次表的績效，若隨機性過大，則可能使原規劃的最佳化排程與班次表失去其最佳性。此等最佳化規劃的排程與班表在營運中受隨機世界擾動的影響，在過去甚少為實務界或學術界所探討。有鑑於此，為能反映真實之市場需求狀況，本研究以航空業者的立場，考量實際營運時市場需求之隨機變動狀況及實務的營運限制，並以業者最大營運利潤為目標，構建一隨機性需求之機隊排程模式。本研究模式期能於未來實務的應用上，提供航空業者一有效的工具，以期幫助航空業者在隨機環境中，規劃最佳的機隊排程與班次表。 本研究利用網路動技巧構建模式，此模式含有多個人流時空網路與機流時空網路，以定式旅客與飛機在時空中的流動情況。在人流時空網路與機流時空網路中及其間，再加上實務的營運限制，以符合實際的飛航作業。此模式可定式為一非線性多重貨物網路流動問題，屬NP-hard問題。本研究利用模擬技巧及數學規劃軟體，發展一以模擬為基礎之啟發解法，以求解模式。為評估隨機性需求排程模式與解法所求得的排程與班次表，在隨機環境中營運的績效，本研究發展一模擬評估方法。最後，為測試本研究模式與解法的實用績效，本研究以一國籍航空公司之國內客運營運資料為例，並利用C電腦程式，配合CPLEX數學規劃軟體進行求解，於個人電腦上進行實例分析，進而提出結論與建議。 The setting of a good flight schedule for an airline not only has to consider its fleet supply, related operations and market share in competition, but also has to consider the uncertainty of the market demand in actual operations. Most of the past research on short-term flight scheduling used the draft timetable and an average passenger demand as input to produce the final timetable and schedule, neglecting the variation of daily passenger demands in actual operations. Although a scheduling model was recently established to deal with variable demands under market competitions, the daily market demand was still assumed fixed in the model, neglecting the stochastic characteristics of daily passenger demands in actual operations, which could affect the optimal fleet routes and timetables. If the market demand is wildly changed in daily operations, then the original schedule could be disturbed to lose its optimality. In the past, the effect of the stochastic disturbance on the optimally planned schedule was rarely researched in either practices or academics. Considering the stochastic characteristics of daily market demands in actual operations, in this research, on the basis of the airline’s perspective, we developed a stochastic-demand scheduling model, with the objective of maximizing the carrier profit, subject to the real operating constraints. The model is expected to be a useful planning tool for airlines to solve their optimal fleet routes and timetables in their short-term operations. We employed network flow techniques to construct the model, which will include multiple passenger and fleet-flow time-space networks in order to formulate the flows of passengers and aircraft in the dimensions of time and space. Some side constraints will be set between the passenger and fleet-flow time-space networks according to the real operating requirements. The model is expectedly formulated as a nonlinear multiple commodity network flow problem that is characterized as an NP-hard problem. We employed simulation techniques, coupled with a mathematical programming solver, to develop a simulation-based heuristic to solve the problem. To evaluate the stochastic-demand scheduling model and the solution method under stochastic demands in actual operations, we developed a simulation-based evaluation method. Finally, to test the model, the solution method and the evaluation method in practice, we will perform a case study on personal computers, using real operating data from a major Taiwan airline, with the assistance of C computer programs and the mathematical programming solver, CPLEX. Conclusions and suggestions will then be given. |
