The Study On Robust Optimization Method For Equitable And Effective Distribution Of Relief Food

The humanitarian relief food distribution problem is that intermediary agencies collect and integrate the donated food from suppliers,as well as distribute it to the people in need.The manner and the amount of distribution are the focus in this thesis.When modeling the relief food distribution problem,there are a lot of uncertainties in the real life,especially uncertain supply,receiving capacity,demand and so on.How to distribute the relief food equitably and effectively in an uncertain environment is our main challenge.To deal with this challenge,we utilize the robust optimization approach to model the uncertain parameters and make effective decisions.Considering that uncertain supply and receiving capacity own interval uncertainties without distribution information,this thesis utilizes the classic robust optimization approach to model the humanitarian relief network to solve the problem of equitable and effective food distribution.Involving the conflicting indexes of equity and effectiveness,this thesis proposes a robust goal programming model for the humanitarian relief food distribution problem.Under the box ? generalized budgeted perturbation set,the model relies on a computationally tractable and parameterized convex programming.We apply the model to a case study on the equitable and effective distribution of food relief in food banks in the United States.In addition to the uncertain supply and capacity,we also take into account the uncertainty of demand.Under the uncertain distribution information,the humanitarian relief food distribution is modeled on the basis of distributionally robust optimization.We thus put forward a distributionally robust model for the equitable and effective food distribution problem.With the mean value and support information,the model depends on a computationally tractable and parameterized conic quadratic programming under the box ? ellipsoidal perturbation set.We apply the model to a case study on the equitable and effective distribution of food relief after an earthquake in Iran.The main new ideas in this thesis are summarized in three aspects as follows: First,this thesis models the equitable and effective distribution of humanitarian relief food.When uncertain parameters are distribution-free,we come up with a novel robust goal programming model considering the classifications of demand points.When uncertain parameters possess partial distribution information,we propose a new distributionally robust model to solve the food distribution problem equitably and effectively.Second,owing to the existence of uncertainty,the proposed models are semi-infinite programming such that the models are difficult to solve.We transform the models into the computationally tractable and parameterized convex programming forms which can be solved directly by software.Third,we verify the practicability and flexibility of the proposed models through the actual case studies.