摘要
为应对新能源日内功率偏差导致的系统电力平衡问题,电力调度部门需要从前瞻性视野对日前计划进行分钟级修正。为此,提出一种基于冗余约束快速辨识的电力系统前瞻调度鲁棒决策方法,在保证短时间高效求解的同时,提升极端运行场景下电力系统的新能源消纳率和保供能力。首先,在前瞻调度决策模式下,考虑风电/光伏的不确定性,结合水电、火电、储能及联络线多资源协同控制建立两阶段优化模型,第1阶段考虑系统的经济性和安全性,第2阶段考虑系统的新能源消纳和保供能力。进一步,考虑到新能源不确定性将增加模型复杂度,提出基于图卷积神经网络-长短期记忆网络的冗余约束辨识与削减方法,并利用列与约束生成算法对约简后的前瞻调度模型进行鲁棒优化求解。最后,在IEEE 118节点系统中进行算例仿真验证,通过数值结果验证了所提模型能够实现冗余约束的快速辨识,所提鲁棒决策方法能对多种不同类型场景的前瞻调度策略进行高效求解。
To address the electric power balance issues in systems caused by intraday power deviations of renewable energies,power dispatch departments require minute-level revisions to the day-ahead plans from a look-ahead perspective.Therefore,a robust decision-making method for the look-ahead dispatch in power system is proposed based on the fast identification of redundant constraints.The method improves the renewable energy accommodation rate and supply capacity of the power system in extreme operation scenarios while ensuring efficient solving in a short time.First,in the look-ahead dispatch decision-making mode,a two-stage optimization model is constructed considering the uncertainty of wind and photovoltaic power,along with the coordinated control of multiple resources such as hydro,thermal,storage,and tie-lines.The first stage considers the economy and safety of the system,while the second stage considers the renewable energy accommodation and power supply guarantee.Furthermore,considering the model complexity added by the uncertainty of renewable energy,an identification and reduction method for redundant constraints based on graph convolution neural network and long short-term memory network is proposed,followed by a robust optimization solution for the reduced look-ahead dispatch model using column and constraint generation algorithm.Finally,the effectiveness of the proposed model for the fast identification of redundant constraints is validated through numerical results in the IEEE 118-bus system,and the proposed robust decision-making method can efficiently solve look-ahead dispatch strategies in various types of scenarios.
作者
王麒宁
陈思远
徐箭
陈亦平
颜融
郭岩
WANG Qining;CHEN Siyuan;XU Jian;CHEN Yiping;YAN Rong;GUO Yan(School of Electrical Engineering and Automation,Wuhan University,Wuhan 430072,China;Southern Power Grid New Energy System(Beijing)Research Institute Co.,Ltd.,Beijing 102218,China;Power Dispatch and Control Center of China Southern Power Grid,Guangzhou 510663,China)
出处
《电力系统自动化》
北大核心
2025年第22期123-134,共12页
Automation of Electric Power Systems
基金
国家重点研发计划资助项目(2022YFB2403500)。
关键词
新能源
决策
极端运行场景
前瞻调度
冗余约束
图卷积神经网络
长短期记忆网络
鲁棒优化
renewable energy
decision-making
extreme operation scenario
look-ahead dispatch
redundant constraint
graph convolution neural network
long short-term memory network
robust optimization
