以取樣方式產生之期望收益函數求解航空網路營收管理問題

Abstract

自從美國航空公司利用營收管理(revenue management)成功提高營收後，營收管理於航空運輸業已是十分普及的技術。基於現今航空軸輻網路的營運型態，營收管理研究重心亦由當初單一航段問題轉向網路型態問題。目前航空網路問題主要以競價法(bid price control)來進行機位控管，過去研究也多利用數學規劃模式中限制式之對偶關係計算競價值。然而，此方式由於須對各時段所剩機位所產生之無限多種可能進行運算，且實際作業上難以做到即時修正，故不管在方法理論及實務運用上，均存在著相當的限制。 本研究以最佳控管機制之動態規劃模式為基礎，建立一函數來近似動態規劃模式中之收益函數，並藉此產生航空網路問題的機位控管決策。首先，以單一航段問題發展本研究方法之模式及演算流程，在小型範例驗證後，再延伸至多航段網路問題中。另外，為求貼近真實世界情況，彌補無法得知各時段所剩機位之各種可能收益值，因此本研究配合發展以抽樣方式產生近似收益函數的方法，並應用於多航段模式之解題概念中。 在兩航段之數值測試範例結果中，可發現本研究所發展的方法不但有效可行，且相較於不執行機位控管的情況下，本研究之方法的確能提供相當接近最佳解的控管結果。預期未來本研究之取樣近似期望收益函數估計法，不但可為網路營收管理問題帶來新的求解概念，且能提供比目前競價法更有效率產生的機位控管決策方法。

Since American Airlines successfully applied revenue management (RM) to raise its revenue, RM has become a common technique in the airline industry. Due to the current hub-and-spoke operation, the focus of the RM research has shifted from the traditional single-leg problem to the network-type problem. It was noticed that, though a mainstream approach for the network RM problem, the bid price control is ill with complicated procedure and heavy computational load. In addition, there are considerable limitations for this approach in terms of both methodological theory and practical application. Based on the dynamic programming (DP) approach, which generates the optimal control policy, this study develops a method that can generate a suitable seat control policy by approximating the expected revenue function in the DP model. First, this study establishes the model and the associated algorithm for the single-leg RM problem. After performing the mathematical analysis, this approach is extended to the network RM problem. Finally, in order deal with the RM problem under the network context, this study adopts the concept of sampling to generate the approximation function. Based on the result in the numerical experiment of a two-leg numerical example, , the validity and effectiveness of the developed method is verified as the generated solution which is pretty close to the optimal solution and significantly better than the case with no RM control. It is believed that this study should serve as an excellent alternative for the current bid price control and provide an inspiring concept for other network-related RM problems.