摘要
随着电子商务的迅速发展,订单的高时效性和商品需求的频繁波动对移动机器人履行系统(RMFS)的订单拣选环节带来巨大挑战。由于RMFS中的货架具有可移动性,在拣选过程中不断优化和调整货架储位是提高RMFS拣选效率的关键途径之一。针对RMFS动态货位分配问题,首先提出一种考虑商品需求实时变动且拣选台任务均衡的动态货位分配策略,构建最大化货位适配值的数学模型,实现货架的合理定位,并运用改进A*算法规划自动导引车(AGV)拣选路径,以避免AGV频繁转弯;其次,针对该问题的动态连续性和解空间巨大的难点,设计一种结合贪心算法和改进自适应遗传算法的混合算法求解该模型,其中引入SoftMax函数用于适应度值的转换,以动态调整交叉变异概率,采用一种基于相似度值的个体交叉判断方法以减少无效交叉,并加入灾变策略防止算法早熟,同时实时更新仓库布局状态以适应每周期的商品需求变化;最后,通过对比不同规模的仿真实验进行验证。结果表明,所提算法能够在保证解质量的同时,大幅度提升求解效率;且随着订单量的增加,所提策略优化效果越明显,即周转率高的货架有更大概率靠近拣选台;相比于静态固定分配策略和动态就近分配策略,所提策略在大规模实验中分别缩短42%和21%的AGV总拣选路径,显著提高了该系统拣货效率。研究结果为RMFS货位分配的进一步优化研究提供参考。
With the rapid development of e-commerce,the high timeliness of orders and frequent demand fluctuations are posing significant challenges to the order picking process of Robotic Mobile Fulfillment Systems.Due to the mobility of pods in RMFS,continuously adjusting pod locations during the picking process is an effective way to improve picking efficiency.This paper proposed a dynamic pod location assignment strategy to address the challenges of dynamic pod allocation in RMFS.The strategy took into account real-time fluctuations in product demand while balancing the workload across picking stations.A mathematical model was constructed to maximize pod allocation value,achieving optimal pod positioning.The improved algorithm was used to plan the picking paths of Automated Guided Vehicles,reducing the number of turns.To address the challenges of dynamic continuity and a vast solution space,a hybrid algorithm combining a greedy algorithm with an improved adaptive genetic algorithm was designed to solve the model.The SoftMax function was introduced to convert fitness values,dynamically adjusting crossover and mutation probabilities.A similarity-based crossover judgment method was employed to reduce ineffective crossovers.A catastrophe strategy was included to prevent premature convergence,and the warehouse layout was updated in real-time to adapt to the changing product demand in each cycle.Finally,the simulation experiments of different scales were compared for verification.The results show that the proposed algorithm can significantly improve solution efficiency while maintaining solution quality.With the increase of order volume,the optimization effect of the proposed strategy becomes more pronounced,with high-turnover pods having a greater probability of being located closer to picking stations.Compared to static fixed allocation and dynamic nearest allocation strategies,the proposed strategy can reduce the total AGV picking paths by 42%and 21%respectively in large-scale experiments,significantly enhancing the system’s picking efficiency.The research results can provide a reference for further optimizing the pod location assignment in RMFS.
作者
田帅辉
林诗阳
TIAN Shuaihui;LIN Shiyang(School of Modern Posts,Chongqing University of Posts and Telecommunications,Chongqing 400065,China)
出处
《铁道科学与工程学报》
北大核心
2025年第5期2089-2099,共11页
Journal of Railway Science and Engineering
基金
重庆市教育委员会人文社会科学研究项目(22SKGH127,23SKGH420)
重庆市社会科学规划项目(2023NDYB79)。
关键词
移动机器人履行系统
自动导引车
动态货位分配
自适应遗传算法
贪心算法
robotic mobile fulfillment systems
automated guided vehicles
dynamic pod location assignment
adaptive genetic algorithm
greedy algorithms
