摘要
风光新能源固有的间歇性和波动性给大规模电力系统发电资源的调度带来了难题,风光储多能互补是应对风光新能源大规模并网的可行途径之一。为制定考虑风光储多能互补的电力系统的优化调度方案,首先,考虑总运行成本最小、新能源弃电量最小的目标,构建了风光储多能互补优化调度模型。然后,面对庞大的新能源规模和逐步完善的电力系统网架结构所带来的优化调度模型求解困难的问题,基于马尔科夫决策过程和近似动态规划理论,将涉及多时段联合求解的优化模型解耦为所有时段单独求解的子问题。在此基础上,采用深度神经网络对解耦后的子问题进行逐时段的求解,提出了一种基于近似动态规划和深度神经网络的深度优化算法。最后,通过在仿真软件和实际大规模电力系统上进行算例测试,验证了所提方法可行性与有效性。
The inherent intermittency and volatility of new energy sources such as wind and solar power present challenges to the scheduling of power generation resources in large-scale power systems.The complementary use of wind,solar,and energy storage is one of the viable strategies to address the large-scale integration of wind and solar energy into the grid.To develop an optimization scheduling plan for a power system that takes into account the complementarity of wind,solar,and energy storage.Firstly,the objective of minimizing total operating costs and the amount of new energy waste is considered.A wind-solar-energy storage multi-energy complementarity optimization scheduling model is constructed.Then,to tackle the difficulty of solving optimization scheduling models due to the vast scale of new energy and the gradually evolving power system grid structure,based on the Markov decision process and approximate dynamic programming theory.It decouples the optimization model,which involves multi-time period joint solving,into sub-problems to be solved separately for each time period.On this basis,deep neural networks are utilized to solve the decoupled sub-problems sequentially,and a deep optimization algorithm based on approximate dynamic programming and deep neural networks is proposed.Finally,the feasibility and effectiveness of the proposed method are validated through numerical tests on both simulated and actual large-scale power systems.
作者
杨银国
谢平平
刘洋
陆秋瑜
徐展鹏
黄泽杰
YANG Yinguo;XIE Pingping;LIU Yang;LU Qiuyu;XU Zhanpeng;HUANG Zejie(Power Dispatching and Control Center of Guangdong Power Grid Co.,Ltd.,Guangzhou 510030,Guangdong,China;China Energy Engineering Group Guangdong Electric Power Design Institute Co.,Ltd.,Guangzhou 510663,Guangdong,China)
出处
《电网与清洁能源》
北大核心
2025年第7期122-131,共10页
Power System and Clean Energy
基金
南方电网公司科技项目(036000KK52222035(GDKJXM20222356))。
关键词
多能互补
风光新能源
输电网
近似动态规划
深度神经网络
multi energy complementarity
wind and solar new energy
transmission network
approximate dynamic programming
deep neural networks
