航空公司為了減少因訂位旅客取消訂位或未報到，所造成對營收之損失，通常都 會採取超額訂位（ overbooking ）的措施。 但若接受的訂位數太多，而在起飛時造成已訂 位旅客無法上機，則航空公司必須因而付出賠償旅客的成本，且可能造成商譽的損失。因此 ，如何決定適當的超額訂位限額是非常重要的課題。 本研究應用多目標數學規劃法所構建之模式，除了可以較有效率地進行超額訂位限額的決定 ，也同時考量營收最大化及商譽損失最小化這二個目標。此外，並探討模糊多目標數學規劃 之目標模糊性問題，不但符合實際應用上的需要，並可滿足不同的決策需求。經本研究之實 例驗證，發現本研究所構建之數學規劃模式，所求解超額訂位限額比人員經驗的決定方式為 佳，而此一模式的求解結果更可幫助提高班機的承載率及營收。

　　　　　A flight may still fly with empty seats even though the booking demand for the flight is higher than its capacity. To overcome this problem, airline companies will usually implement an overbooking policy. However, overbooking will result in "deny boarding", and then lead to the "penalty cost" and loss of the company's reputation. How to decide a proper booking limit is thus a very important issue on yield management. To help airline companies efficiently and effectively decide an optimal booking limit, this research develops a multi-objective model and a Fuzzy multiobjective model to achieve both "maximum revenue" and "minimum loss of goodwill". This study may not only fit with practical needs in actual applications, but also satisfy various needs of decision-making. The results of this study suggest that using the model developed to determine optimal booking limits compare favorably with the rule-of-thumb approach in terms of the load factor and revenue of flights.