基于风险规避航空货运舱位优化控制RAMDP模型-《计算机技术与发展》

文章信息/Info

Keywords:: air cargo; inventory control; risk aversion; Markov decision processes

摘要:: 航空货运订舱存在不确定性,航空货运收益管理决策者不仅要考虑期望收益最大化,还要考虑决策风险,需要通过风险规避降低风险损失.针对不确定性的舱位优化控制问题,利用马尔可夫决策方法,结合期望效用理论的冯.诺依曼-摩根斯坦效用函数,构建了基于风险规避的有限阶段马尔可夫决策RAMDP模型并给出了相应的算法过程.与基于目标期望收益最大化的模型相比,模型不仅具有结构化舱位保护策略,而且通过调整风险规避因子可为决策者提供不同的决策选择.算例分析说明了该模型的适用性

Abstract:: With uncertainty of air cargo booking,air cargo revenue decision-makers don’t only consider maximizing expected revenue, but also consider the decision-making risk,policy makers need risk aversion to reduce the risk loss. With Markov decision-making meth-od and von Neumann-Morgenstern utility function of expected utility theory,a finite stage Markov decision processes model based on risk aversion,RAMDP for short,is established for the air cargo uncertain inventory control problem,also,the corresponding algorithm is giv-en. Compared with the model based on the target expected revenue maximization,the model has not only structured protection strategy, but also through the adjustment of risk aversion factor,can provide decision makers with different decision-making choices. A numerical example is given to illustrate the application o f the proposed model