基于深度学习的水风光短期随机优化调度研究
Short-term Stochastic Optimization Scheduling of Hydro-Wind-Solar Power System based on Deep Learning
张一凡1
作者信息
- 1. 三峡大学电气与新能源学院,湖北 宜昌 443002
- 折叠
摘要
我国致力于可再生能源发展,提出水-风-光多能互补系统,因风光能源的不确定性,需实时电网调度调整.文章运用深度学习(DQN)优化系统的短期调度,最大化发电效益.采用拉丁超立方抽样和考虑Kantorovich距离的场景削减技术,反映可再生能源不确定性分布,结合深度强化学习建立多能互补系统短期优化调度模型.模拟实际数据,显示该方法有效解决高维等问题,较于传统方法有显著优势.
Abstract
Renewable energy is developing rapidly in China.A hydro-wind-solar power complementary system is thus proposed.While,due to the uncertainty of the wind and solar powers,real-time scheduling of the power grid is required.The deep Q network(DQN)is used to optimize the short-term scheduling of the system and maximize the power genera-tion efficiency.The Latin hypercube sampling and the scenario reduction technique considering the Kantorovich distance are used to reflect the uncertainty distributions of the renewable energies.Then,a short-term optimization scheduling model is developed for the multi-energy complementary system with the deep reinforcement learning.Simulation with ac-tual data shows that the proposed method can solve high-dimensional problems efficiently and has significant advantages over traditional methods.
关键词
短期调度/不确定性/拉丁超立方抽样/场景削减/Deep/Q/NetworkKey words
short-term scheduling/uncertainty/Latin hypercube sampling/scenario reduction/deep Q network引用本文复制引用
出版年
2024
NETL
NSTL
万方数据