随机需求多车辆路径问题的神经网络算法

A neural network algorithm for multiple vehicles routing problem with stochastic demand

目前国内外对随机需求多车辆路径问题的研究还很少,本文针对标准hopfield神经网络容易陷入局部极值点等问题,以总路程最短和总使用车辆数最少为目标,提出了一种基于退火策略的混沌神经网络的求解随机需求多车辆路径问题的算法,该算法既可以使混沌运动有足够长的进程以提高粗搜索性能,又可以随混沌动态的减弱使收敛速度加快。实验结果表明,该算法优化车辆路径更佳,是解决随机需求多车辆路径问题的有效方法。

In recent times,the research of multiple vehicles routing problem（VRP） with stochastic demand is not consummate in domestic and abroad.However,this paper indicates a kind of annealing chaotic neural network to multiple vehicles routing problem with stochastic demand,targeting to the shortest total distance and the least number of vehicles,while standard Hopfield neural network easy to fall into local optimum.In this algorithm,chaos lasts long enough to improve the search performance of crude,and also convergence rate is larger while chaotic dynamics weakens.According to the experimental results,vehicle routing is better using this algorithm,the algorithm is a effective method to solving multiple vehicles routing problem with stochastic demand.