請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/86215
標題: | 混合接單與服務期並考慮升級之飯店預訂控制 Hotel Booking Control with Mixed Booking-Service Periods and Upgrades |
作者: | Ting Lin 林亭 |
指導教授: | 孔令傑(Ling-Chieh Kung) 孔令傑(Ling-Chieh Kung | lckung@ntu.edu.tw | ), |
關鍵字: | 飯店預訂控制問題,多服務期,動態規劃,啟發式演算法,需求不確定性, hotel booking control,multiple service periods,dynamic programming,myopic policies,demand uncertainty, |
出版年 : | 2022 |
學位: | 碩士 |
摘要: | 預訂控制為收益管理的主要策略之一,特別是對於飯店業、航空業等出租高價值資產以賺取利潤的行業。基於這些行業供給固定、服務 具有時效性以及未來需求具不確定性的特性,以預訂控制來挑選更具價值的消費者訂單是一項重要的收益管理策略,同時也能有效地應對因供給彈性不足而造成供需不匹配的問題。在這些行業中,飯店業由於可以連續訂數天的房間,在進行預訂控制時需同時考慮不同天的空房數與隨機需求,使其預訂控制尤其複雜。 本文探討考慮升級之多房型、多服務期的飯店預訂控制問題。在我們的問題中,旅行社會一次預訂多日、多種房型的多個房間並要求折扣;而散客則被視為每個需考慮的服務期間收到的隨機需求,不會要求折扣。本研究主要目標為對於旅行社進行適當的預訂控制策略,以最大化飯店之預期收益。 基於此問題需要一連串的決策,且每一期的決策會影響到未來所有決策,我們提出了動態規劃模型以求此問題的最佳解。模型中訂單與服務可能發生在同一期,且同時要考慮所有可能的升級方案。然而因多日多種房型所導致的高複雜度,使得該動態規劃模型無法在實務上可接受的時間內找到最佳解。因此我們提出一啟發式演算法,以便能較有效率地求得近似解,並透過數值實驗檢驗此演算法的效能,以提供在不同情境下的預訂控制決策建議。 Booking control is an essential strategy in revenue management for the industries which provide perishable products for rental, such as the hotel and airlines industries. For the hotel industry, the problem is especially complicate since an order may request a room for multiple days, which means the decisions on a single order needs to consider the capacity and demand uncertainty of multiple days at the same time. In this study, we consider a hotel booking control problem with multiple room types, multiple service periods, and multiple upgrade plans. Booking and service may exist together in a single period, which is the main difference from previous studies. In the problem, decisions on order acceptance and upgrade plan need to be made. Our objective is to propose a way to make such decisions to maximize the total expected revenue. To maximize the total expected revenue, a dynamic programming (DP) model is formulated to find the optimal decisions on the acceptance and upgrade plan of the orders received from travel agencies. Due to the high dimensionality of states, solving the DP model is too time-consuming, which have us to propose a myopic policy that makes decisions fast enough for practical use. The numerical experiments show that myopic policy is near-optimal and very time efficient on this problem, especially when there are lots of customers and orders. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/86215 |
DOI: | 10.6342/NTU202202221 |
全文授權: | 同意授權(全球公開) |
電子全文公開日期: | 2022-09-12 |
顯示於系所單位: | 資訊管理學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
U0001-0908202218081300.pdf | 1.67 MB | Adobe PDF | 檢視/開啟 |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。