计算机与现代化 ›› 2025, Vol. 0 ›› Issue (06): 106-113.doi: 10.3969/j.issn.1006-2475.2025.06.017

• 信息系统 • 上一篇    下一篇

基于深度强化学习的居民微电网调度算法

  

  1. (1.南京南瑞继保电气有限公司,江苏 南京 211100; 2. 国家电网内蒙古东部电力有限公司电力科学研究院,内蒙古 呼和浩特 010000)
  • 出版日期:2025-06-30 发布日期:2025-07-01
  • 作者简介: 作者简介:郭鑫溢(1989—),男,宁夏石嘴山人,工程师,硕士研究生,研究方向:新能源,储能发电并网控制技术,E-mail: guoxy@ nrec.com;姜凯(1989—),男,江苏盐城人,工程师,研究方向:电力系统监控,新能源数据分析,E-mail: jiangkai@nrec.com; 杨朋威(1989—),男,河北邢台人,工程师,研究方向:电力系统及其自动化,E-mail: yangpengwei@md.sgcc.com.cn; 陈更(1995—),男,内蒙古赤峰人,工程师,研究方向:继电保护及自动化,E-mail: chengeng99999@163.com。
  • 基金资助:
    基金项目:内蒙古自治区科技重大专项资助项目(2021ZD0039)

Residential Microgrid Scheduling Algorithm Based on Deep Reinforcement Learning

  1. (1. Nanjing Nari Relay Electrical Co., Ltd., Nanjing 211100, China;
    2. State Grid Inner Mongolia Eastern Electric Power Co., Ltd., Hohhot 010000, China)
  • Online:2025-06-30 Published:2025-07-01

摘要:
摘要:居民微电网调度是在应对高度波动的用电需求下,寻求电网最优运行状态的复杂任务。由于系统的复杂性、不确定性和动态性,需依赖高效、精确的决策方法。本文提出基于深度Q学习(DQN)的调度模型——DQN for Load Scheduling(DQN-LS),可精准调控不同时间和节点的负荷,兼顾线路容量、发电能力、节点需求等多重约束,并考虑季节、时间、天气等因素,实现快速准确决策。为评估其性能,本文将DQN-LS与融合CNN与LSTM的Agent模型、蒙特卡洛树搜索(MCTS)、多智能体软演员评论(MASAC)等主流算法对比。实验结果表明,DQN-LS在平均奖励、奖励稳定性、调度效率及约束违例等方面均优于对比模型,验证了其在居民微电网调度中的有效性与优势。


关键词: 关键词:居民微电网, 电力负荷, 调度决策, 深度强化学习

Abstract:
Abstract: Residential microgrid scheduling is a complex task aimed at optimizing grid operation under highly variable power demands. Due to the system's complexity, uncertainty, and dynamics, efficient and accurate decision-making is essential. This paper proposes a DQN-based model, DQN for Load Scheduling (DQN-LS), which accurately schedules residential loads across different times and nodes while considering constraints such as line capacity, generation limits, and node demand. It also factors in seasonal, temporal, and weather variations to ensure fast and precise decisions. To evaluate its performance, DQN-LS is compared with leading methods, including CNN-LSTM-based agents, Monte Carlo Tree Search (MCTS), and Multi-Agent Soft Actor-Critic (MASAC). Experimental results, using real datasets and simulations, demonstrate that DQN-LS outperforms baseline models in average reward, reward variance, scheduling efficiency, and constraint violations, confirming its effectiveness and superiority in residential microgrid scheduling.

Key words: Key words: residential microgrid, electrical load, scheduling decision, deep reinforcement learning

中图分类号: