一类随机需求VRP的混合粒子群算法研究

针对一类随机需求车辆路径问题(stochastic vehicle routing problem,SVRP),结合现实生活中长期客户服务记录所隐含的统计性知识构建新的统计学模型,并将种群搜索与轨迹搜索算法相结合提出了一种新的混合粒子群优化算法。该算法通过引入导引式局部搜索,来减小粒子群搜索陷入局优的可能性以获得更优化解。仿真计算证明混合粒子群优化算法的有效性。同时,该算法也拓展了VRP的算法空间。

Hybrid particle swarm optimization algorithm for stochastic vehicle routing problem

To solve the stochastic vehicle routing problem,a novel algorithm,i.e hybrid particle swarm optimization(H-PSO),which combines the population search and the path search algorithms,is proposed based on the new statistical model constructed with the statistical knowledge obtained from the real-life long-term customer service records.This algorithm introduces the guided local search to reduce the passibility that the search for particle swarp falls into the local optima so as to obtain a more optimal solution.The simulation test proves the validity of the H-PSO,and this algorithm also extends the algorithm of VRP.