Research On Robust Optimization Of Multistage Inventory Problem With Uncertain Demand

With the development of times and technology,people are facing more intense market competition.How to enhance competitiveness in market competition is a strategic issue that companies first consider.In the process of enterprise survival and development,logistics cost has always been a key factor in which inventory management is a link that cannot be ignored.This paper starts with the inventory problem in real life,and establishes a multistage inventory management model with uncertain demand from the perspective of uncertain market demand,which results in the uncertain demand in each stage.In order to satisfy all inventory constraints,a series of robust optimization methods are used to solve the problem.This article first analyzes the traditional inventory system and model,introduces some classic inventory management strategies,and then proposes a multistage inventory management model with uncertain demand.This paper briefly introduces the idea and theory of traditional robust optimization method and its applications in inventory models.Starting from the model,this paper analyzes the theory of the adjustable robust optimization method and the Affinely Adjustable Robust Counterpart for multistage problem.Aiming at the situation that the real data is not timely and inexact,this paper proposes Weakly Affinely Adjustable Robust Counterpart method and Inexact Affinely Adjustable Robust Counterpart method: while affinely adjustable robust method is highly dependent on the known data,weak method improves the affine decision rules so that it can still be effectively used when the information is incomplete;for the inexactness of the real data and the infeasibility of the affinely adjustable robust method in this case,the disturbance of the data is added to the inventory model to improve the robustness of the solution.This article conducts experiments with randomly generated demand data.The improvement of traditional robust optimization by affinely adjustable robust counterpart under standard conditions is verified.The dependence of affinely adjustable robust counterpart method on known data is illustrated by examples,and the effectiveness of weak method is verified.Finally,through a random example,the infeasibility of the affinely adjustable robust optimization in the case of inexact data is illustrated,and at the same time,the validity of the inexact method is verified.