Logistics Service Composition Based on Multi-Objective Optimization

Given the reform in railway freight transportation, the authors developed an end-to-end logistics service to respond more promptly and effectively to customers’ diverse, individual needs. They built a Web-based service model, incorporating all the characteristics of traditional logistics services. This Web-based model dynamically composed the services offered by logistics providers in real time and transformed them into multi-objective optimization problems. Then the authors applied an improved NSGA-Ⅱ（non-dominated sorting in genetic algorithm-Ⅱ） algorithm to the optimization problem. Firstly, in the chromosome encoding and genetic factor of the algorithm, they considered the logical relationship of the logistic services and the associations between the logistics service provider alliances. Secondly, they added customers’ demand for the QoS （quality of service） attribute as a penalty function to the fitness function, effectively avoiding the problem of service re-planning and encouraging the evolution of the chromosomes in the direction of the constraint. The experimental simulation of the algorithm shows that it is a feasible and effective solution to the logistics service composition problem. In an actual railway logistics service, the QoS attribute of specific and composite services meets customer requirements and the composite service is non-inferior for all objectives. The running time of the algorithm meets customer needs even when the number of logistics tasks and candidate services increases. The proposed algorithm offers a better logistics service composition plan compared to the NSGA-Ⅱ algorithm, because the Pareto solution that meets customers’ needs accounts for 100% of the Pareto solution set and also considers the demands of the allied logistics service providers.