多级正向变异的遗传算法提高求解VRP问题效率

应用遗传算法对车辆路径问题(VRP)求解时,由于遗传算法在解决VRP问题时,交叉操作难以保留优秀基因片段,可能导致算法收敛较慢等问题.在一定程度上影响了遗传算法解决VRP问题的实用性.在前人的基础上,通过一种多级正向变异方法,使变异最大程度向好的方向进行,拆除基因片段中较差的基因连接并建立新基因连接,从而得到较优的新基因片段,重复一定的变异次数,让变异达到最优效果.通过实验表明多级正向变异明显提高了遗传算法解决此类问题的效率.

Multi-level forward mutation genetic algorithm to improve the efficiency for VRP

When using genetic algorithm to solve VRP problem, a slower convergence problem may be generated because the crossover operation could not keep good genes, which affects the usefulness of genetic algorithms to solve the VRP problem to a certain extent. On the basis of our predecessors, we have created a multi-level forward mutation method which dismantles poor gene fragment connection and creates a new connection to get a better gene. A large number of experiments show that forward mutation can greatly improve the genetic algorithm to solve such problems efficiently.