數字孿生輔助的變電站巡檢任務分配算法
張樂霏
(三峽大學 電氣與新能源學院，湖北 宜昌 443002)
摘 要 ：變電站巡檢自動化、智能化需求增長，但現有的四足巡檢機器人面臨能耗、電池與充電問題， 影響巡檢效率。對此提出了數字孿生輔助的多目標強化學習算法 (DT-PPO)，構建場景感知能耗模型量化動態 功耗，并引入動態權重遷移機制自適應調整策略。實驗結果顯示，相比傳統方法，該算法顯著提升任務完成率、 電池利用率，降低充電頻次，表明DT-PPO在復雜動態環境下具有魯棒性與實用性，實現了任務分配與能量 管理協同優化。
關鍵詞 ： 數字孿生 ；強化學習 ；能量管理 ；多場景任務分配 ；變電站巡檢
中圖分類號 ：TM63 文獻標識碼 ：A 文章編號 ：1007-3175(2025)11-0035-06
Digital Twin-Assisted Substation Inspection Task Allocation Algorithm
ZHANG Le-fei
（College of Electrical Engineering & New Energy, China Three Gorges University, Yichang 443002, China）
Abstract: With the increasing demand for automation and intelligence in substation inspections, existing quadruped inspection robots face problems such as energy consumption, battery and charging issues, which affect the inspection efficiency. To address this, a digital twin-assisted multi-objective reinforcement learning algorithm (DT-PPO) is proposed. This algorithm constructs a scenario-aware energy consumption model to quantify dynamic power consumption and introduces a dynamic weight transfer mechanism to adaptively adjust strategies. Experimental results show that compared with traditional methods, this algorithm significantly improves task completion rate and battery utilization and reducing charging frequency. These results indicate that DT-PPO has robustness and practicality in complex dynamic environments, achieving collaborative optimization of task allocation and energy management.
Key words: digital twin; reinforcement learning; energy management; multi-scene task allocation; substation inspection
參考文獻
[1] HUTTER M, GEHRING C, LAUBER A, et al.ANYmaltoward legged robots for harsh environments[J]. Advanced Robotics，2017，31(17)：918-931.
[2] CHEN X, LIU Y, ZHANG Q.Energy management strategies for quadruped robots in dynamic environments[J].IEEE Transactions on Robotics， 2021，37(4)：1125-1137.
[3] 周立輝，張永生，孫勇，等 . 智能變電站巡檢機器人研制及應用[J]. 電力系統自動化，2011，35(19)： 85-88.
[4] MA Y, CHEN H, WANG Z.Dynamic power consumption analysis for legged robots on variable terrains[J].IEEE Transactions on Robotics， 2017，33(4)：872-885.
[5] TAWFIK M, PATTERSON A, SMITH J.Nonlinear energy characterization in multi-task robotic operations[J].Journal of Field Robotics，2024， 41(1)：45-62.
[6] MOHAJER N, ABBASI R, MARTINS L.Quantitative analysis of lithium-ion battery degradation in robotic applications[J].Energy Storage Materials，2020，28： 132-144.
[7] LIU X, ZHANG K, WU Y. Peak-valle gap amplification caused by fixed-threshold charging in microgrids[J].Applied Energy，2022， 308：118324.
[8] JUAN E, GARCI A S, MARTI NEZ P.Real-world performance degradation of offline-trained RL agents in mobile robotics[J].Robotics and Autonomous Systems，2023，159：104288.
[9] ZHANG L, WANG H, CHEN Y.Digital twin-enhanced reinforcement learning for multi-objective optimization in smart grids[J].IEEE Transactions on Industrial Informatics，2023，19(5)：4321-4332.
[10] LI X, LIU S.Dynamic battery management in robotic systems using adaptive threshold control[J].Energy Conversion and Management， 2022，265：115741.
[11] 盧成，袁明輝 . 鋰離子電池容量動態測量研究 [J]. 上 海理工大學學報，2013，35(2)：165-168.
[12] SHANNON C E. A mathematical theory of communication[J].The Bell System Technical Journal，1948，27(3)：379-423.
[13] 王清宇 . 基于電池老化的并聯式混合動力汽車SOC估算及能量管理策略研究 [D]. 重慶：重慶交通大學，2021.
[14] 馮光.基于EKF的鋰離子電池SOC估算的建模與仿真[D]. 武漢：武漢理工大學，2013.
[15] 周秀文 . 電動汽車鋰離子電池健康狀態估計及壽命預測方法研究 [D]. 吉林：吉林大學，2016.
