计及碳捕集装置及碳排放潮流理论的电力系统低碳优化学习调度  

Low Carbon Optimal Learning Scheduling for Power Systems With Carbon Catchment Devices and Carbon Flow Theory

作　　者：李吉峰 邹楠 李卫东[2] 张明泽 吴俊 LI Jifeng;ZOU Nan;LI Weidong;ZHANG Mingze;WU Jun(State Grid Liaoning Electric Power Supply Co.,Ltd.,Dalian Electric Power Supply Company,Dalian 116001,Liaoning Province,China;School of Electrical Engineering,Dalian University of Technology,Dalian 116024,Liaoning Province,China)

机构地区：[1]国网辽宁省电力有限公司大连供电公司,辽宁省大连市116001 [2]大连理工大学电气工程学院,辽宁省大连市116024

出　　处：《现代电力》2024年第5期854-865,共12页Modern Electric Power

摘　　要：针对当前电力调度方面的研究未将碳排放潮流与电力潮流进行协调融合,以及求解算法的智能性仍需进一步挖掘等问题,提出计及碳捕集装置及碳排放潮流理论的电力系统低碳优化学习调度方法。首先,分别从设备层面与系统层面构建电力系统碳排放潮流模型;其次,综合考虑电力系统内源–网–荷–储各个环节,建立包含系统日前调度与负荷需求响应调整的双层交替优化调度模型,并采用深度强化学习算法对模型进行求解;最后,通过实际算例仿真验证所提理论方法在降低运行成本与碳排放量方面的有效性与适用性。In allusion to the problems that the current research on power scheduling does not integrate the carbon emission flow with the power flow as well as the intelligence of the solution algorithm still needs to be explored,a low-carbon optimal learning and scheduling method of power systems that took into account the carbon capture device and the carbon emission flow theory was proposed.Firstly,the power system’s carbon emission flow model was constructed at the equipment and the system level respectively.Secondly,a bi-level alternating optimal scheduling model,which includes system day-ahead scheduling and load demand response adjustment,was established by considering each link of source-grid-load-storage of the power system,and a deep reinforcement learning algorithm was adopted to solve the model.Finally,the effectiveness and applicability of the proposed theoretical approach in reducing operating costs and carbon emissions were verified through actual example simulations.

关 键 词：碳捕集 碳排放潮流 运行调度 深度强化学习 需求响应 

分 类 号：TM73[电气工程—电力系统及自动化]