Application Of Genetic Algorithms In Logistics System Optimization

Logistics has been considered to be the "third source of profit" after the reduction of the consumption of raw material and improvement of productivity. By optimizing logistics system can reduce the cost of logistics system and improve competitiveness of corporations. So it is very significant to make researches on optimizations in logistics system.Inventory cost and distribution cost are core cost of logistics system; they take high percentage of total cost of logistics system. So it is able to reduce the cost of logistics system effectively if inventory cost and distribution cost can be reduced.Genetic Algorithm(GA) has been widely studied, experimented and applied in many fields. GA is analysed and studied in this paper, some improved methods are presented accroding to the shortages of GA. Based on the above research, inventory optimization and vehicle routing problems in logistics are studied in the paper. Inventory simulation optimization problem and vehicle routing problem are regard as combinatorial optimization problem and are solved by GA in this paper.The main achievements of the dissertation are as follow:(1) Simulation models of inventory systems are built based on decrete event system. One of difficulties of applying system simulation to obtain the optimial control parameters of the inventory systems is simulation optimization. GA is adopted as the algorithm of simulation optimization in the paper. A candidate policies' collector which can collect pareto solutions and M elites selection operator are designed due to randomicity of stochastic inventory system. Simulation results of two typical inventoy system show the simulation optimization methods presented in the paper is feasible and effective.(2) Traveling Salesman Problem(TSP) is a sub-problem of Vehicle Routing Problem(VRP). In order to solve TSP, three crossover operators which are common used in solving TSP are studied firstly. Then, an efficient hybrid genetic algorithms for TSP(HGA-TSP) is presented in this paper. HGA-TSP adopts a variational OX operator as crossover operator and adopts 2-opt as mutate operator. K Nearest point Sets is also proposed to decrease the search space.(3) Biobjective multiple travelling salesman problem(MTSP) model is built forvehicle routing problem with single distribution, multiple vehicles and none of restriction. Two objectives are total distance and the equilibrium of each subtour. Three solutions to biobjective MTSP are also studied based HGA-TSP algorithm in this paper.(4) A new Double Layers Chromosome Coding Schema(DLCCS) and a new Subtour Exchange Algorithm(SEA) are proposed for Capacitated Vehicle Routing problem(CVRP). DLCCS assure each subroutine is feasible and the optimal vehicle number of the CVRP is needless when encoding because each chromosome is only composed from the customers number in DLCCS. so DLCCS is more suitable to solve the practical VRP whose minimal vehicle number is unknown in advance; besides DLCCS can reduce search space and improve search efficiency. SEA is able to improve the precision of GA. In order to solve CVRP, two hybrid genetic algorithms, HGA-CVRP and HGA-SEA-CVRP are proposed in the paper, HGA-CVRP is based on DLCCS but HGA-SEA-CVRP combines HGA-CVRP with SEA.(5) A hybrid genetic algorithm for vehicle routing problem with time windows(HGA-VRPTW) is also proposed. In order to handle subtours more conveniently, DLCCS is improved firstly In HGA-VRPTW. After that, how combine GA and local search algorithm is studied and add a local search operator with clone operation to GA.(6) Vehicle Routing Problem Simulation Lab(VRPSL) is developed to study vehicle routing problem. VRPSL is a software package with graphic user interface and easy to use. It plays an important role in research process of this paper.Lots of benchmarks are computed in the paper. Computing results show the algorithms presented in this paper are able to solve inventory optimization and Vehicle Routing Problem effectively.