In recent years,along with the frequent occurrence of large-scale emergencies,the demand of emergency logistics has become more urgent.Suddenness of emergency logistics,the weak economy,and the uncertainty makes it largely different from normal logistics.In order to cope with emergency logistics of these characteristics,relevant models and algorithms need to be designed.Multi-depot vehicle routing problem(MDVRP)is an extension of the traditional vehicle routing problem.The characteristics of multi-supply make it more accord with the actual situation of emergency logistics.Bashed on the standard MDVRP model,a model which contains supply and demand was proposed in this paper.The model set the minimum and maximum demand to simulate the uncertain demand as well as the imbalance of supply and demand when emergency logistics occurs.Taking into account the vehicle routing problem is a typical NP-Hard problem,this paper designed two heuristic algorithms to be able to obtain a better solution at the occurrence of emergency logistics.First,through the use of boundary coefficient,demand points are divided into different groups in the improved genetic algorithm.Multi-depot problem is transformed into several single-depot problems.Through the different selection of boundary coefficient,a large number of initial solution were generated by saving algorithm and scanning algorithm,which ensures the diversity and effectiveness of the initial population.An improved crossover operator is applied in this algorithm,which can maintain more line information while reducing the damage to the existing lines in respect to the existing operators.Through the integration of improved genetic algorithm and tabu search algorithms,this paper implements the memetic algorithm.By using a standard test set of MDVRP,we found that the improved genetic algorithm mentioned above has a great advantage on the traditional genetic algorithm.At the same time,we also found that the memetic algorithm can provide a much better solution.Finally,a visual emergency logistics VRP software was developed by using SuperMap GIS and Java.The front end was developed by Flash,which makes the software have strong cross-platform features.Different models were contained in the back-end server to achieve the emergency logistics VRP optimization decision support. |