智能电网中基于Q学习的能量双边拍卖算法  

Energy Double Auction Algorithm for Smart Grid Based on Q Learning

作　　者：刘迪迪[1] 张泉景 邹艳丽[1] 秦运柏[1] 孙浩天 胡聪[2] LIU Didi;ZHANG Quanjing;ZOU Yanli;QIN Yunbai;SUN Haotian;HU Cong(College of Electronic Engineering,Guangxi Normal University,Guilin,Guangxi 541004,China;Guangxi Key Laboratory of Automatic Detecting Technology and Instruments,Guilin University of Electronic Technology,Guilin,Guangxi 541004,China)

机构地区：[1]广西师范大学电子工程学院,广西桂林541004 [2]广西自动检测技术与仪器重点实验室(桂林电子科技大学),广西桂林541004

出　　处：《南方电网技术》2021年第7期109-115,共7页Southern Power System Technology

基　　金：国家自然科学基金(62061006);广西自然科学基金资助项目(2018JJA170167);广西科技基地和人才专项(2018AD19342);广西创新驱动发展专项(桂科AA21077015);广西自动检测技术与仪器重点实验室开放基金项目(YQ18202)。

摘　　要：针对智能电网中多个参与电力市场交易的终端用户,首先通过引入双边拍卖机制,构造了一个包含多用户的能量交易市场模型,然后将设计的多用户交易模型转化成信息不完全的非合作博弈模型,为使多个用户间能量交易趋于稳定,基于Q学习创新地提出了一种自适应学习算法,为参与博弈的用户找到最优混合策略,并且使整体博弈达到混合策略纳什均衡,从而使多用户能量交易能够稳定运行,最后通过数值仿真生成混合策略纳什均衡的策略概率分布,证实了提出的多用户能量交易算法的有效性。Considering multiple end-users taking part in the electric power market transaction of the smart grid,an energy trading market model of multiple end-users is constructed by introducing the double auction mechanism in this paper firstly.Then the model is transformed into a non-cooperative game model with incomplete information.To make energy trading among multiple users tend to be stable,an adaptive learning algorithm is proposed for the end-users playing the game to find the optimal mixed-strategy based on the Q learning,and the overall game comes up to Nash equilibrium,so that energy trading between the multiple end-users can run stably.Finally,the probability distribution of mixed-strategy for multiple end-users to obtain Nash equilibrium under numerical simulation,the validity of the proposed energy trading algorithm is verified.

关 键 词：能量拍卖 非合作博弈 双边拍卖 Q学习 智能电网 

分 类 号：TM73[电气工程—电力系统及自动化]