| 引用本文: | 周雪松,韩 静,马幼捷,等.基于DQN算法的直流微电网负载接口变换器自抗扰控制策略[J].电力系统保护与控制,2025,53(1):95-103.[点击复制] |
| ZHOU Xuesong,HAN Jing,MA Youjie,et al.Active disturbance rejection control strategy of a DC microgrid load interface converter based on a DQN algorithm[J].Power System Protection and Control,2025,53(1):95-103[点击复制] |
|
| 摘要: |
| 在直流微电网中，为了保证直流母线与负载之间能量流动的稳定性，解决在能量流动中不确定因素产生的扰动问题。在建立DC-DC变换器数学模型的基础上，设计了一种基于深度强化学习的DC-DC变换器自抗扰控制策略。利用线性扩张观测器对总扰动的估计补偿和线性误差反馈控制特性对自抗扰控制器结构进行简化设计，并结合深度强化学习对其控制器参数进行在线优化。根据不同工况下的负载侧电压波形，分析了DC-DC变换器在该控制策略、线性自抗扰控制与比例积分控制下的稳定性、抗扰性和鲁棒性，验证了该控制策略的正确性和有效性。最后，在参数摄动下进行了蒙特卡洛实验，仿真结果表明该控制策略具有较好的鲁棒性。 |
| 关键词: 直流微电网 深度强化学习 DQN算法 DC-DC变换器 线性自抗扰控制 |
| DOI：10.19783/j.cnki.pspc.240480 |
| 投稿时间：2024-04-21修订日期：2024-06-03 |
| 基金项目:国家自然科学基金重点项目资助(U23B20142)；天津市研究生科研创新项目资助(2022BKYZ036) |
|
| Active disturbance rejection control strategy of a DC microgrid load interface converter based on a DQN algorithm |
| ZHOU Xuesong1,HAN Jing1,MA Youjie1,TAO Long1,WEN Hulong2,ZHAO Ming3 |
| (1. Tianjin University of Technology, Tianjin 300384, China; 2. Tianjin Ruineng Electric Co., Ltd., Tianjin 300381, China;
3. Chengde Dianzhishang Energy Saving Technology Co., Ltd., Chengde 067000, China) |
| Abstract: |
| In the DC microgrid, to ensure the stability of the energy flow between the DC bus and the load, the disturbance problem caused by the uncertain factors in the energy flow is solved. Based on the mathematical model of DC-DC converter, an active disturbance rejection control strategy for a DC-DC converter based on deep reinforcement learning is designed. The active disturbance rejection control structure is simplified using the estimation compensation of the total disturbance and the linear error feedback control characteristics of the linear expansion observer, and the controller parameters are optimized online by deep reinforcement learning. From the load-side voltage waveform in different working conditions, the stability, immunity and robustness of the DC-DC converter using the control strategy, linear active disturbance rejection control and proportional integral control are analyzed, and the correctness and effectiveness of the control strategy are verified. Finally, Monte Carlo experiments are carried out under parameter perturbation, and the simulation results show that the control strategy has good robustness. |
| Key words: DC microgrid deep reinforcement learning deep-Q-network (DQN) algorithm DC-DC converters linear active disturbance rejection control |
