Particle swarm optimization for open vehicle routing problem in dynamic network

Amiming at the dynamic and open constraints of the existing vehicle routing problem, the mathematical model of the open Vehicle Routing Problem (VRP) in dynamic network was established. And the time dependent function was used to represent dynamic network. Particle swarm optimization with self-adaptive inertia weight and classified status update was proposed to solve the problem. According to social cognitive theory, each particle regulated its inertia weight dynamically according to the relative value of the particle's current position with its best position in the history and the best position in the population. To avoid premature convergence, classified update strategies were used to increase population diversity. For the excellent particles, their information entropy was computed after server iterations and their position were updated. For the inferior particles updates, it were conducted by recording frequencies in the board and displaced by the new particles. In the experiment, the parameters were analyzed. Comparing to other algorithms on benchmarks showed that the algorithm was effective.