面向大型工业仓库的多AGV综合优化调度方法  

An integrated scheduling method of multi-AGV in large-scale industrial warehouses

作　　者：胡恩泽 贺建军[1] 申帅 吴仁超[1] HU Enze;HE Jianjun;SHEN Shuai;WU Renchao(School of Automation,Central South University,Changsha 410083,China)

机构地区：[1]中南大学自动化学院,湖南长沙410083

出　　处：《中南大学学报（自然科学版）》2023年第5期1779-1790,共12页Journal of Central South University:Science and Technology

基　　金：国家自然科学基金资助项目(61873282)。

摘　　要：针对大型工业仓库中自动引导车(AGV)集群执行大规模任务时效率低的问题,提出一种基于分层规划的综合优化调度方法,将调度问题分解为聚合的上层任务定序分配问题和下层路径规划问题。在上层问题中生成精英解集,将下层路径规划生成的禁忌列表作为碰撞冲突约束条件融入上层问题迭代寻优过程,并在迭代搜索中,将路径规划结果反馈给上层问题并更新约束条件,通过上下两层问题的综合计算得到整体性能最优的方案。进一步将路径搜索和禁忌列表融合,实现AGV调度中相互关联的多个问题的同步优化。通过大型工业仓库实例验证该方法的有效性。研究结果表明:相较于依次优化调度方法,基于分层规划的综合优化调度方法的平均任务完成时间减少10.56%,碰撞冲突造成的延迟时间减少74.53%;相较于混合自适应大邻域搜索算法和预先规划算法,基于混合离散状态转移算法的分层规划综合调度方法对于大规模问题任务完成时间分别减少9.73%和5.54%,计算时间分别减少84.19%和86.68%。To address the inefficiency of automated guided vehicle(AGV) fleets in large-scale industrial warehouses,an integrated scheduling method based on hierarchical planning was proposed,which decomposed the scheduling problem into an aggregated upper-level task sequencing assignment problem and a lower-level path planning problem.An elite solution set was generated in the upper-level problem,and the tabu list generated by the lower-level path planning was integrated into the iteration search process as collision conflict constraints for upperlevel problem.In iteration search processes,the elite solution of the upper-level problem with a large number of collision conflicts obtained by the path planning was continuously excluded by updating the upper-level elite solution set and the lower-level tabu list,to obtain the solution with the best overall performance.Further integrating path search and tabu lists to achieve simultaneous optimization of multiple interrelated problems in AGV scheduling,the effectiveness of the method was verified by a large-scale industrial warehouse case.The results show that,compared to the sequential optimization scheduling method,the completed time and delay time due to collision conflicts of an integrated scheduling method reduce 10.56% and 74.53%,respectivelty.For largescale problems,compared to the hybrid adaptive large neighborhood search algorithm and the pre-planning algorithm,the task completion time of an integrated scheduling method reduce by 9.73% and 5.54%,respectively,and the computation time reduce by 84.19% and 86.68%,respectively.

关 键 词：自动引导车(AGV) 综合优化调度 任务定序分配 路径规划 碰撞冲突 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]