论文部分内容阅读
竞价控制是收益管理中广泛应用的一种存量控制方法.将网络存量控制问题描述为一个动态规划模型,通过状态向量的一个仿射函数近似动态规划的最优值函数,并且在航段水平上考虑随机需求,最终得到一个计算网络竞价所需的确定性线性规划(DLP),相对于标准的DLP,这个DLP得到了更接近于动态规划最优值的上界.给出了一个列生成算法用于求解这个DLP,并提供了模拟算例,计算结果表明可获得比标准的DLP方法更好的收益.
Bidding control is a stock control method widely used in revenue management.The network stock control problem is described as a dynamic programming model which approximates the optimal value function of dynamic programming through an affine function of the state vector and at the segment level Considering the stochastic demand, we finally get a deterministic linear programming (DLP) which is required to calculate the network bidding. Compared with the standard DLP, this DLP gets an upper bound that is closer to the optimal value of the dynamic programming. A column generation algorithm It is used to solve this DLP and provides a simulation example. The calculation results show that a better profit can be obtained than the standard DLP method.