Multi-tour Optimization Policy for Stochastic Vehicle Routing Problem
随机车辆路径问题的多回路优化策略

Stochastic demands enhance complexity and difficulty of decision-making in the process of vehicle routing.Assumed that exact demands of customers are obtained only after vehicle visit them,and can not be divided,a version of vehicle routing problem with stochastic customers and stochastic demands(VRPSCD) is introduced.Firstly,multi-tour policy is put forward,and its asymptotic property is analyzed.To find a superior prior tour,several simulated annealing algorithms with different neighborhood structures are designed.Experiments demonstrate validity of multi-tour policy,and show superiority of the simulated annealing with combined neighborhood.