基于粒子群算法的航班舱位控制研究

Airline Seat Inventory Control Based on Particle Swarm Optimization

首先介绍了收益管理提出的背景思想，在对航班收益管理基本概念作简要阐述的情况下，给出了期望边际座位收入理论．提出了在北京到上海的单航程航段下，航空公司会提供出发时间不同但航程相同的连续两个航班．旅客在没能购买到期望的航班舱位时，会以一定的概率选择购买下一航班的机票或者取消订票，改乘其他交通运输工具．旅客对于某一航班的座位需求主要由固有需求和上一航班需求的转移两部分构成．基于这样的角度建立了两航班机票预售模型，不考虑动态订座，no-show和超售等情况，并利用算法对模型进行了求解和分析．最后得出结论当航班票价等级数量较少时，为高票价等级安排较多数量座位可提高航班收入．而当航班票价等级数量比较多时，为低票价等级多安排一些座位可为航班提高收入．

The article firstly introduces the background knowledge of revenue management In the case of briefly describes the basic concepts, it gives the marginal seat revenue expectations theory. This paper presents a case in a single leg from Beijing to Shanghai. The airline will provide flights which departure continuously. Guests who can＇t attain desired flight seat will choose to buy the next flight tickets or cancel bookings, switch to other means of transporta- tion. The demand of a certain flight is composed of two parts, the one is the inherent demand and the other one is transfer needs. Based on such an angle, the article established the two flights of revenue management model. It does not consider dynamic reservation, no-show and overbooking, etc. Last, it uses algorithms to solve the model and gives the results, when the number of fare classes is small, a large number of seats for high fare classes can improve airline revenue, when the number of fare classes is more, it is more appropriate to arrange more seats for the low fare levees.