电工技术学报  2020, Vol. 35 Issue (13): 2757-2768    DOI: 10.19595/j.cnki.1000-6753.tces.191746
多种可再生能源互补发电系统的规划与运行专题(特约主编:陈哲教授 胡维昊教授) |
基于深度强化学习算法的光伏-抽蓄互补系统智能调度
李涛1, 胡维昊1, 李坚1, 韩晓言2, 陈哲3
1.电子科技大学机械与电气工程学院 成都 611731
2.国家电网四川省电力公司 成都 611041
3.奥尔堡大学能源系 奥尔堡 DK-9110
Intelligent Economic Dispatch for PV-PHS Integrated System: a Deep Reinforcement Learning-Based Approach
Li Tao1, Hu Weihao1, Li Jian1, Han Xiaoyan2, Chen Zhe3
1. School of Mechanical and Electrical Engineering University of Electronic Science and Technology of China Chengdu 611731 China
2. State Grid Sichuan Electric Power Company Chengdu 611041 China
3. Department of Energy Technology Aalborg University Pontoppidanstraede 111 Aalborg Denmark
全文: PDF (1531 KB)   HTML
输出: BibTeX | EndNote (RIS)      
摘要 抑制高渗透率并网光伏发电的功率波动对电力系统稳定及经济运行影响至关重要。储能系统通过充放电行为有效抑制光伏发电功率的波动,从而保障光伏电源平滑接入大电网。针对现有光伏发电功率预测精度不高,储能设备工况转换频繁导致系统稳定性降低以及经济收益差等问题,该文以100%消纳光伏发电为前提,提出采用周期衰减学习率的改进型深度确定性策略梯度算法(CDLR-DDPG)的光伏-抽水蓄能互补发电系统的实时智能调度方法;通过将智能调度问题转换为马尔可夫决策过程并对其求解,得到抽蓄的实时运行策略。以四川省小金县某光伏电站的历史数据为例,对所述方法进行了仿真。结果表明,该文所提出的方法能有效地缓解并网点功率波动,减小抽蓄工况转换频率,提升光伏并网电力系统的稳定性和经济性,在运行周期内使得并网点功率波动降低了12.7%,同时互补系统收益增长了4.95%。
服务
把本文推荐给朋友
加入我的书架
加入引用管理器
E-mail Alert
RSS
作者相关文章
李涛
胡维昊
李坚
韩晓言
陈哲
关键词 光伏抽水蓄能互补发电智能调度深度强化学习    
Abstract:It is important for power system stability and economic operation to effectively suppress the power fluctuation from the large-scale grid-connected photovoltaic (PV) stations on the point of common coupling (PCC). Energy storage system (ESS) may effectively provide the power support to smooth the PV output power to the main grid. However, there are some drawbacks for PV power forecast, such as low accuracy, the frequently charging and discharging exchange of ESS, which may result in low stability and economic benefits. Based on this motivation, the modified cycling decay learning rate- deep deterministic policy gradient (CDLR-DDPG) approach is proposed in this paper, to implement the online intelligent economic dispatch for the PV-PHS complementary power generation system considering both the power fluctuation on PCC and the promise of full absorption of PV. The Markov decision process is introduce to convert this dispatch model and the CDLR-DDPG algorithm is adopted to solve it. Finally, a case study is carried out to evaluate the performance of intelligent economic dispatch model based on the real PV plant obtained in Xiaojin County, Sichuan province, China. The simulation results reveal that the intelligent dispatch strategy can effectively mitigate the power fluctuation and enhance the economic efficiency, that is, the power fluctuation on PCC is reduced by 12.7% and the economic revenue of complementary system is increased by 4.95%, simultaneously.
Key wordsPhotovoltaic    pump hydro-storage    hybrid energy    intelligent dispatch    deep reinforcement learning   
收稿日期: 2019-12-13     
PACS: TM615  
基金资助:国家重点研发计划资助项目(2018YFB0905200)
通讯作者: 胡维昊 男,1982年生,教授,博士生导师,研究方向为人工智能在电力系统中的应用、可再生能源发电技术。E-mail:whu@uestc.edu.cn   
作者简介: 李 涛 男,1994年生,博士研究生,研究方向为可再生能源与储能系统分析与智能调度。E-mail:tli@std.uestc.edu.cn
引用本文:   
李涛, 胡维昊, 李坚, 韩晓言, 陈哲. 基于深度强化学习算法的光伏-抽蓄互补系统智能调度[J]. 电工技术学报, 2020, 35(13): 2757-2768. Li Tao, Hu Weihao, Li Jian, Han Xiaoyan, Chen Zhe. Intelligent Economic Dispatch for PV-PHS Integrated System: a Deep Reinforcement Learning-Based Approach. Transactions of China Electrotechnical Society, 2020, 35(13): 2757-2768.
链接本文:  
https://dgjsxb.ces-transaction.com/CN/10.19595/j.cnki.1000-6753.tces.191746          https://dgjsxb.ces-transaction.com/CN/Y2020/V35/I13/2757