航空運量預測與航空網路設計之研究-應用灰色理論

Abstract

航空運量預測、航空網路形態設計、航線班機頻次規劃與機型指派等工作 ，為航空公司短期運作規劃，如排班、人員調度之前期規劃，其結果直接 影響航空公司之營運效益與服務品質。過去航空運量預測相關研究大多應 用多元迴歸模式、時間序列模式等，均須要求大樣本量或較好之分佈規律 。另一方面，航空網路、班機頻次規劃與機型指派問題上之相關研究，則 大多為確定性之解析模式及數學規劃最適法，在這些研究中，對於處理不 確定性參數環境，尚存有極大之探討空間。灰色理論，本質上所須數據量 少，且具適用於處理不確定性參數之特性，故本研究以一航空公司客運網 路為對象，系統化應用灰色理論進行一系列研究，構建航線運量預測、航 空網路形態設計、航線班機頻次規劃與機型指派模式。本研究第一部份為 航空運量預測模式之構建，應用並修正灰色預測理論構建航線運量預測模 式，並以包絡灰區間估計未來運量發展的可能範圍，以作為研擬航線形態 設計、航線頻次規劃與機型指派之主要輸入參數。第二部份為航線形態之 設計，考慮航線長度指標、中途停靠站數目、航線旅客運量集中程度等因 素，並對這些因素之不確定性，應用灰色聚類法進行路線候選，以設計航 線網路形態。第三部份規劃航線頻次與機型指派，以追求網路航線中航空 公司營運成本與旅客成本最小為目標，考慮航線頻次規劃與機型指派對航 空公司營運成本與旅客延誤時間之間的權衡取捨，對未來運量之動態變化 ，進行動態規劃。最後，並以一我國國籍航空公司 ( C航) 為對象，進行 實證範例分析。研究結果顯示，運量預測模式對運量變化的擬合與上下包 絡灰區間對運量未來發展的捕捉，均較傳統迴歸模式與時間序列模式具有 較佳之能力；而航線網路形態設計結果與 C航實際現況比較，結果合理； 航線班機頻次規劃與機型指派模式在機隊航機數目限制下，規劃結果為指 派各航線最適機型與最適頻次，相較 C航現有班次時刻表，在配合未來運 量成長與動態變化之下，規劃結果良好，驗證本研究模式確實可行，而配 合動態規劃模式之執行結果，將可提供航空公司在規劃未來航空網路，更 高之決策彈性與參考價值。本研究成果不僅在學術上可作為航空運量預測 、航空網路設計等相關研究之參考，以及航線排班、機隊指派等研究之依 據。所發展之模式亦可提供航空公司在航空網路前期規劃上之應用與參考 基礎。 Airline traffic forecast and network design including the decisions of a network shape, flight frequencies and aircraft types on individual routes are prerequisite to air carrier's operational planning such as scheduling and crew assignment. The chosen network shape, flight frequency and the aircraft type on individual routes directly affect the operating effectiveness of the air carrier and the quality of service provided to passengers. Grey prediction and decision-making theory, in contrast to conventional methods, is a convenient mathematical device of which to forecast airline traffic with small amount of data and to solve problems containing uncertainty, and indetermination. This research develops a series of models to forecast airline city-pair passenger traffic, to design a network of airline routes, to determine flight frequencies and to assign aircraft on individual routes by applying Grey prediction and decision-making theory. In the first part of the study, we develop an improved Grey prediction model to forecast airline's city-pair traffic and to estimate upper and lower limits to capture the extents of variations in future trends. These forecasted values are used as input data of the other two models. The second part of the study defines the route length index, the total number of intermediate stops and the concentration index of traffic flow and uses these three indexes to determine the route candidates by applying Grey clustering. The third part, based on Grey programming and mathematical programming, determines flights' frequencies and assigns aircraft on the route candidates that minimize the total airline operation and passenger travel cost. Finally, a case study with an airline passenger network (China airlines) and the other data is provided to illustrate the results and the application of the models. We find that the forecasting models developed do produce less prediction error. The results of the case study verify that the models are practicable, and also provide higher flexibility on decision-making for airlines. This study demonstrates how Grey theory might be applied to the air traffic forecasting and the airline network design problems. In additions, it is envisaged that the results of this study may shed light on strategic and operational planning related issues for air carriers.