Solving Facility Location Problems Based On Meta-heuristic Algorithms

The facility location problem(FLP)is the basic decision of the logistics industry and the supply chain industry.It is also a crucial decision.The quality of the location decision will not only directly affect the construction cost,transportation cost,storage cost,the efficiency of cargo distribution,and the overall benefits,but also indirectly affect the entire logistics chain or other links in the supply chain.Therefore,the correct location decision has important significance for the upstream and downstream industries of the logistics industry and even the economy and society.The facility location problem is an important research area of operations research.Because of its academic value and application value,this problem has attracted the attention of many researchers and published many research results.Today,as a huge research area,facility location has developed many subfields and had fruitful research results.The research content of the topic is to design metaheuristic algorithms to solve the facility location problem.This dissertation is focused on the reliable facility location problem,and based on the research of reliable facility location problem,it is further extended to the study of the robust and reliable facility location problem.Based on the extensive literature review,we constructed a corresponding appropriate mathematical model for the problem.Combining the local search method with the framework of a kind of classic metaheuristic algorithms,evolutionary algorithms,we proposed some metaheuristic algorithms: Evolutionary Algorithms with Weak Local Search(EAWLS),Evolutionary Algorithms with Strong Local Search(EASLS),and Evolutionary Algorithm with Memorable Local Search(EAMLS)to solve the model.For the two research problems(reliable facility location problem and robust and reliable facility location problem),we have constructed two different models for comparison and designed detailed experiments for each model.We compared our algorithm with the Lagrangian relaxation algorithm and one of the world's most advanced optimization solvers,IBM CPLEX,on different scale problem instances to verify the effectiveness of our algorithm.In addition,we developed a web demo based on the Django framework to solve the facility location problem instances,and implemented some basic functions.On the webpage,there is a basic introduction about the facility location problem,the mathematical models we built,and the EAMLS algorithm,and it can receive parameters entered by the user,generate a corresponding problem instance,call the EAMLS algorithm to solve the instance and show the location decision,location diagram and algorithm convergence curve diagram.The innovation of our research is reflected in the following aspects.(1)We construct a new reliable facility location problem model which adopts a new allocation method;(2)The scale of large-scale problems we solve(600-node)is larger than other literature(maximum 150-node);(3)A robust and reliable facility location problem model is proposed based on the above-mentioned RFLP model by scenarios planning method;(4)A novel hybrid evolutionary algorithm is proposed which performs good on small,mid,and large-scale problems;(5)A population diversity metric 0-HDR and a convergence metric l3-value are proposed to help users observe the evolutionary process,adapt parameters,and analyze and improve the algorithm;(6)A simple web application demo is developed for facility location problem,which has some foundational functions.