Research On E-commerce Logistics Distribution Based On Hybrid Genetic Algorithm

In traditional logistics distribution,there are some problems such as low degree of automation,delay of information and bad connection of network. All of these affect the efficiency of logistics distribution and its development. In a networked environment, advanced techniques combined with logistics distribution can shorten the delivery,improve the service quality and increase the competitiveness of enterprises. This kind of novel model has many advantages that can improve the efficiency of logistics distributionThis thesis focused on the distribution routing optimization problem occurred in the B2 C electronic environment. In this environment, the routing problem has some new characters, such as the large number of customers, dispersed customers, as well as rigorous requirement on service time, compared with the traditional vehicle routing problem. Thus new models and algorithms for this kind of problems should be studied deeply. With regard to these new characters, a new mathematical model with the objective function to find the shortest delivery route and considering the constraint of vehicle loading capacity and running distance was established. A new hybrid algorithm(called CWGA) which combined the genetic algorithm with the saving algorithm was proposed to our optimization problem. Randomly generated instances are used to do computer simulation. For each instance, different experiments are designed according to the principle of the design of experiment and enough simulation experiments are run. Experimental results are statistically studied and comparison analysis is provided. The result shows that the method can improve the logistics distribution's efficiency, and can find a better route than that by traditional genetic algorithm as well by saving method. With the increase of evolution generation, both GA and CWGA have chance to find a good route, but CWGA can use less time to find a better solution because CW algorithm provides a better ‘seed' for GA. Thus CWGA is much suitable for large scale VRP instances. Both evolution generation and population size have influence on the algorithm performance. Because of capacity constraints both of the loading and running length of vehicle, the illegal route is different.The study results can improve the efficiency of logistics distribution, shorten the delivery distance, save the cost of distribution, and improved the service customers. Therefore, the study is both of some theoretical and practical significance.