摘要
为提高可再生能源消纳能力,减少整流和并网等设备的投资成本,降低电解水制氢系统成本,实现可再生能源大规模制氢,构建了一个孤岛型可再生能源大规模制氢系统。该系统通过智慧能量管理,实现了提高系统经济性与安全性的目标。首先建立可再生能源大规模制氢系统的仿真模型,制定控制策略;其次,提出一种基于深度确定性策略梯度(DDPG)的能量优化调度策略。通过大量长期的训练,使用DDPG算法得到的智能体能够实现智能化的动态能量优化调度。将该策略与深度Q网络、粒子群优化和传统控制方法在经济性和安全性方面进行比较,结果表明DDPG算法在能量优化管理中可实现更高的经济收益,更好地利用可再生资源,并确保系统的安全运行。
To improve the renewable energy consumption,reduce the investment on rectifiers and grid connection equipment,cut down the cost of water electrolysis for hydrogen production through powering hydrogen production by renewable energy,an islanded renewable energy large-scale hydrogen production system is constructed.An intelligent energy management platform can improve the economy and safety of the system.Firstly,a simulation model of the renewable energy large-scale hydrogen production system is established and its control strategy is formulated.Secondly,an energy optimization scheduling strategy based on deep deterministic policy gradient(DDPG)algorithm is proposed.Through long-term trainings,the agent obtained from the DDPG algorithm can achieve intelligent dynamic optimized scheduling on energy.Comparing the performances of the proposed strategy with deep Q network(DQN),Particle Swarm Optimization(PSO)and traditional control methods in terms of economy and safety,it is shown that applying the DDPG algorithm in energy optimization and management can get higher economic returns and utilization rates of renewable resources,and ensure the safe operation of the system.
作者
郑庆明
井延伟
梁涛
柴露露
吕梁年
ZHENG Qingming;JING Yanwei;LIANG Tao;CHAI Lulu;LYU Liangnian(Hebei Jiantou New Energy Company Limited,Shijiazhuang 050011,China;School of Artificial Intelligence,Hebei University of Technology,Tianjin 300401,China;Goldwind Science&Technology Company Limited,Beijing 102600,China)
出处
《综合智慧能源》
CAS
2024年第6期35-43,共9页
Integrated Intelligent Energy
基金
河北省科技支撑计划项目(F2021202022)
国家重点研发计划项目(2023YFB3407703)。
关键词
可再生能源
大规模制氢
离网型
深度确定性策略梯度
优化调度
renewable energy
large-scale hydrogen production
off-grid
deep deterministic policy gradient
optimized scheduling
