Joint Location-inventory Model Based On K-means Clustering Genetic Algorithm

Logistics distribution center is one of the major facilities in logistics system. The location of distribution center is of vital importance. Location problem of distribution center is the most active and valuable problem in logistics theory research. In recent years, more and more scholars take inventory policy into account when researching on location problem. A series of joint location-inventory models have been built. Deep research on joint location-inventory model has important theoretical significance and practical significance.Based on previous research and a specific problem, this paper builds a joint location-inventory model for a type of product whose demand rate is time-varying and influenced by inventory level. And then author extends the model to another two models which considers the case that time-varying-demand product and stochastic-demand product exists meantime. The first model of the two has an Objective of minimizing the cycle costs. Objective of the second is minimizing the cycle costs and maximizing the time satisfaction at the same time.To the models established, this paper gives a hybrid algorithm named k-means clustering genetic algorithm which is based on k-means clustering algorithm and genetic algorithm. It takes four steps:(1)cluster retailers by using k-means clustering algorithm;(2) judge candidate location point by evaluation function value and obtain candidate location decisions;(3) find the best allocation decision under a certain candidate location decision by using genetic algorithm;(4) compare the value of objective function and find the optimal solution.With MATLAB, author writes the program of k-means clustering genetic algorithm. By solving a specific problem, author proves that the models established are practical and meaningful and the algorithm developed is useful for solving discrete location problem. Compared with Enumeration Method and Greedy Dropping Heuristic Algorithm, solution obtained by the algorithm in this paper saves money in a large percentage.