当前位置: X-MOL 学术Comput. Ind. Eng. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Modelling data-driven distributionally robust risk-averse hub interdiction median problem under hypothesis test
Computers & Industrial Engineering ( IF 7.9 ) Pub Date : 2021-04-16 , DOI: 10.1016/j.cie.2021.107323
Fanghao Yin , Yi Zhao

This paper explores a hub interdiction median problem for risk-averse managers using the objective function of the total travel time. This problem mainly involves the selection of the key hub with the strongest influence on the hub network and the redesign of this network after disruptions. Motivated by the mean and conditional value-at-risk criteria (CVaR), we first study the maximum total travel time in the hub network after disruptions and propose a mean-CVaR hub interdiction median model. Then, we consider the travel time in the model as a random variable with finite sample observations. Assuming that the probability distribution of random time has the known finite discrete support, we develop a novel data-driven distributionally robust mean-CVaR hub interdiction median model by constructing an ambiguity set based on the robust optimization method and the Pearson chi-square hypothesis test. The constructed ambiguity set includes the real probability distribution of the travel time at a designated confidence level. We further derive the worst mean via robust sample average approximation approach, and obtain a data-driven cone representation of the worst CVaR under ambiguity set. Finally, by using the travel time data from a local aviation hub network in the United States, we conduct the numerical experiments in a real-world context to demonstrate the validity of our proposed model.



中文翻译:

在假设检验下对数据驱动的分布鲁棒的风险厌恶中心拦截中位数问题进行建模

本文利用总旅行时间的目标函数,为规避风险的管理者探索了中枢拦截的中位数问题。该问题主要涉及选择对集线器网络影响最大的关键集线器,以及在中断后重新设计该网络。受均值和条件风险值标准(CVaR)的推动,我们首先研究了中断后枢纽网络中的最大总旅行时间,并提出了均值CVaR枢纽拦截中位数模型。然后,我们将模型中的旅行时间视为具有有限样本观测值的随机变量。假设随机时间的概率分布具有已知的有限离散支持,通过构建基于鲁棒优化方法和Pearson卡方假设检验的歧义集,我们开发了一种新的数据驱动的分布式鲁棒均值CVaR枢纽枢纽拦截中位数模型。构造的歧义集包括在指定的置信度下的行进时间的实际概率分布。我们通过鲁棒的样本平均逼近方法进一步得出最差均值,并在歧义集下获得最差CVaR的数据驱动锥体表示。最后,通过使用来自美国本地航空枢纽网络的旅行时间数据,我们在现实世界中进行了数值实验,以证明我们提出的模型的有效性。构造的歧义集包括在指定的置信度下的行进时间的实际概率分布。我们通过鲁棒的样本平均逼近方法进一步得出最差均值,并在歧义集下获得最差CVaR的数据驱动锥体表示。最后,通过使用来自美国本地航空枢纽网络的旅行时间数据,我们在现实世界中进行了数值实验,以证明我们提出的模型的有效性。构造的歧义集包括在指定的置信度下的行进时间的实际概率分布。我们通过鲁棒的样本平均逼近方法进一步得出最差均值,并在歧义集下获得最差CVaR的数据驱动锥体表示。最后,通过使用来自美国本地航空枢纽网络的旅行时间数据,我们在现实世界中进行了数值实验,以证明我们提出的模型的有效性。

更新日期:2021-04-30
down
wechat
bug