面向集合任务的多无人机电力巡检任务分配方法研究  

Research on task allocation method for multiple drones power inspection for collective tasks

作　　者：赵慧敏 罗贺[1,4,5] 阴酉龙[2] 林世忠 王国强 ZHAO Huimin;LUO He;YIN Youlong;LIN Shizhong;WANG Guoqiang(School of Management,Hefei University of Technology,Hefei 230009,China;Management Center of UAV Inspection Operation,State Grid Anhui Electric Power Co.,Ltd.,Hefei 230061,China;Anhui Power Transmission&Transformation Engineering Co.,Ltd.,Hefei 230071,China;Ministry of Education Key Laboratory of Process Optimization and Intelligent Decision-Making,Hefei 230009,China;Engineering Research Center for Intelligent Management of Aerospace System of Anhui Province,Hefei 230009,China)

机构地区：[1]合肥工业大学管理学院,合肥230009 [2]国网安徽省电力有限公司无人机巡检作业管理中心,合肥230061 [3]安徽送变电工程有限公司,合肥230071 [4]过程优化与智能决策教育部重点实验室,合肥230009 [5]安徽省空天系统智能管理工程研究中心,合肥230009

出　　处：《系统工程理论与实践》2025年第2期666-684,共19页Systems Engineering-Theory & Practice

基　　金：国家自然科学基金面上项目(71971075,72271076,71871079);国家自然科学基金基础科学中心项目(72188101);国家重点研发计划项目(2019YFE0110300);安徽省博士后研究人员科研活动经费(2022B587)。

摘　　要：在无人机电力巡检过程中,一个待巡检部件通常对应多个位置不同且均符合拍摄要求的任务点,这些拍摄任务点构成一个集合任务.为了保证巡检质量,通常要求无人机多次采集待巡检部件的图片,即访问集合任务中的多个拍摄任务点.本文针对上述特点,将面向集合任务的多无人机电力巡检任务分配问题建模为最小最大化多站点家庭旅行商问题(minmax multi-depot family traveling salesman problem,Minmax-MDFTSP),并设计了一种强化遗传算法框架.在该框架下,提出了染色体校验及修正机制、组合交换变异算子、基于贪婪策略的局部调优算子,设计了基于强化学习的遗传算法参数调优方法.性能实验结果表明,本文方法在求解质量和求解效率方面均具有明显优势.此外,通过消融实验验证了强化遗传算法框架中各个部分的有效性.最后结合实际场景下的具体案例,通过实地飞行验证了本文方法相对于现有巡检方式的优势.During the process of using drones for power inspection,a component to be inspected usually corresponds to multiple task points that are different in location but all meet the photography requirements.These task points form a collective task.To ensure the quality of inspection,it is required for the drone to take multiple shots of the components to be inspected,that is,to visit multiple task points in the collective task.In light of the above characteristics,the problem of task allocation for multiple drones power inspection for collective tasks was modelled as minmax multi-depot family traveling salesman problem(Minmax-MDFTSP).A framework that combines reinforcement learning and genetic algorithm was designed to solve the problem.This framework contained a mechanism for checking and correcting chromosomes,a combination exchange mutation operator,a local optimization operator based on a greedy strategy and a parameter tuning method for genetic algorithm based on reinforcement learning.The results of the performance experiment showed that the proposed method in this paper exhibited significant improvements in both solution quality and solving efficiency.Besides,the ablation experiment confirmed the effectiveness of each part in the framework.Finally,in combination with real-world scenarios,the advantages of the proposed method over existing inspection methods were verified through on-site flight validation.

关 键 词：电力巡检 多无人机 任务分配 强化学习 遗传算法 

分 类 号：V279[航空宇航科学与技术—飞行器设计] TM75[电气工程—电力系统及自动化] TP18[自动化与计算机技术—控制理论与控制工程]