Hub Location Problems With Differentiated Service Levels Under Mixture Uncertain Situations

Hub location problems have always been the focus and hot spot in the optimization field.Reasonable logistics network can greatly reduce the transportation cost,and then bring huge economic benefits to logistics companies.Recently,as China's national economy changes from “seller's market” to “buyer's market”,the customer's position is becoming more and more important.Considering that customers often measure the quality of their service according to the delivery time and the freight of their goods,to improve customer satisfaction,logistics companies should not only ensure that the goods are delivered on time,but also minimize their operating costs to reduce the freight.In view of this,on the promise of meeting the customer's delivery demand,this thesis studies the hub location problem with the goal of minimizing the cost of logistics network.After that,according to the characteristics of the customer delivery demand and the environment of the logistics network,two kinds of research problems in stochastic environment and mixed uncertain environment are discussed,respectively,and then,two kinds of high precision and efficiency algorithms are designed accordingly.The main contents of this thesis are as follows:· Firstly,modeling and processing of hub location problems considering differentiated service levels under single and mixed uncertain environments.In the logistics network,the delivery demand of customers(for example,the goods are expected to be delivered within 24 or 48 hours),is usually considered as the service level to be provided by each logistics company,while the existing research on service level is essentially a single service level.In view of this,this thesis assumes each logistics outlet providing a variety of service levels to provide alternative arrival services for customers.Considering the complex and variable environment of the logistics network,by identifying uncertain variables,this thesis studies the hub location problems with differentiated service levels under single and mixed uncertain environments,respectively.After that,the corresponding mathematical models for these two kinds of location problems are established,and the nonlinear models are linearized according to the characteristics of the logistics network and some mathematical methods.· Secondly,designing algorithms for the hub location problems under two uncertain conditions.The hub location problem in practice involves many logistics outlets,and this thesis considers extra differentiated service levels and uncertain environment,which limits the problem scale of solution software and existing solution methods.In view of this,this thesis focuses on the algorithm design for solving large-scale practical problems.Aiming at the location problem in the case of single uncertainty,This thesis fuses interval reduction algorithm,single allocation principle(a variant of the nearest allocation principle)and meta-heuristic algorithm(including cross entropy algorithm,genetic algorithm and tabu search algorithm)to form 18 hybrid intelligent algorithms to solve this problem.Two-stage methods based on fuzzy simulation algorithm and fuzzy operational law are designed,to solve the location problem under mixed uncertainties.In the two-stage method based on the fuzzy simulation algorithm,this thesis improves the traditional stochastic discretization algorithm according to the fuzzy operational law,and then proposes the uniform discretization algorithm and the improved uniform discretization algorithm,and thus forms four kinds of two-stage methods to solve the problem.· Finally,analysing the experimental results based on Turkish data set.In this thesis,data sets with problem size of 10?50 is firstly generated from the internationally recognized Turkish data set.On this basis,the effectiveness of the two programming algorithms is verified from the algorithm stability,accuracy and efficiency,to find the best algorithms for the two problems.The experimental results show that in terms of algorithms under a single uncertain environment,compared with the other 17 algorithms,a hybrid intelligent algorithm composed of the cross entropy algorithm,an improved single allocation principle and interval reduction algorithm can not only guarantee the error to be less than 1% in most cases,but also can output the result of the problem size of 50 within 500 seconds,which saves nearly 25% to the sub-optimal algorithm.For the algorithms under mixed uncertainty environments,the algorithm precision of the two stage methods based on the improved uniform discretization algorithm and the fuzzy operational law,are roughly similar,with the error of no more than 1% in solving the problem size of 10?50,and their efficiency is roughly same too,both of which can output the results of problem size of 50 less than 80 seconds,while the shortest running time of the other two algorithms is around 10000 seconds.In addition,the results of the two research problems are compared,to illustrate the necessity of considering the mixed uncertain environment.The comparison results show that in the problem size of 50,the result error of the mixed uncertain environment model is basically less than 1%,and its credibility value is also above 90%,while the result error of the single uncertain environment model is up to 5%,and its credibility value is basically floating around 40%.In general,the research results of this thesis,not only enrich the research area of hub location problems,but also provide a new way of thinking for their solving.