改进的Hopfield神经网络在配送车辆调度中的应用研究.pdfVIP

摘 要 本论文通过对物流配送中心运输车辆调度问题的研究，提出了基于模拟 退火算法的混合Hopfield 神经网络算法（SA—HNN ），并用此混合算法求解 带宽时间窗有容量约束配送车辆调度问题。模拟退火算法及Hopfield 神经网 络算法存在着收敛速度慢、易陷入局部最优等一些缺陷，使得系统解决车辆 调度优化问题不令人满意。本文提出了混合神经网络算法（SA-HNN ），此算 法通过结合Hopfield 神经网络与模拟退火算法各自的优点来克服彼此的不 足，将模拟退火中以一定概率接收较差解的机制应用到Hopfield 神经网络算 法中，以克服陷入局部最优的缺陷。同时又对传统的网络和模拟退火算法进 行了改进，最终得到了一种快速、高效、收敛的算法。最后利用提出的混合 算法解决雨润集团的物流配送实际问题，取得了良好的效果。 关键词：Hopfield 神经网络，模拟退火，物流配送，车辆调度，全局优化 Abstract This thesis have researched on the scheduling problem of the vehicles of the logistics distribution center, proposed the blending hopfield neural network algorithm based on simulated annealing algorithm, solved vehicles routing problem with width time windows and the capacity to restrain. The simulation annealing algorithm and the hopfield neural network algorithm has some flaws ,the slow convergence rate, easy to fall into partially most excellent, causes the system not to be satisfying. This article proposed the blending network algorithm, this algorithm unifies respective merit to overcome each others insufficiency through blending the hopfield neural network and the simulation annealing algorithm, using certain probability receive bad solution mechanism in the simulation annealing algorithm to the hopfield neural network algorithm, overcomes the flaws of falling into partially most excellent. Simultaneously has made the improvement to the traditional network and the simulation annealing algorithm, obtained one kind of fast and highly effective restraining algorithm finally. Finally using the mix algorithm to sovle actual distribution problem in Rain Run group, has made the good progress. Key word: Simulation annealing, Hopfield ne