中文摘要：

针对销售系统不能完全记录顾客到达情况，难以准确收集和分析顾客选择偏好，存在较大需求预测误差，使得基于选择行为的存量控制模型在应用中往往达不到理论预期效果的问题，为了系统地减少最优控制策略对需求预测误差的敏感性，应用马尔可夫决策过程理论和稳健最优化方法建立存量控制的稳健模型，并同时考虑超订问题；在证明最优值函数单调的基础上，构建了一种稳健竞标价格策略．数值模拟算例表明，在需求预测误差较大情况下，最优稳健策略比一般最优策略更能够提高总收益．

英文摘要：

Incorporating consumer choice behavior into the process of capacity control decision represents one of the inevitable trends in perishable product revenue management. Because the sales system cannot record the complete data of consumer arrival and exactly collect and analyze the information of consumers＇ preference, demand forecast is usually far from accurate. Inevitable forecast error is an important limiting factor in applying the choice-based capacity control model to industry practice. In order to systematically mitigate the sensitivity of the optimal control strategy to forecast errors, a robust capacity control model was proposed by applying the theory of Markov decision processes and robust optimization approach, which also consider the demands of overbooking. The monotone property of the robust optimal value was proved. On this basis, a robust optimal control strategy, the robust bid-price strategy, was given. Numerical example was given to illustrate that the optimal robust strategy can significantly improve the total revenue in the case of large errors of demand forecast.