基于PPO算法的电热氢耦合综合能源系统优化调度  

OPTIMIZATION AND SCHEDULING OF ELECTROTHERMAL HYDROGEN COUPLED COMPREHENSIVE ENERGY SYSTEM BASED ON PPO ALGORITHM

作　　者：梁涛 张晓婵 谭建鑫[2] 井延伟 吕梁年 Liang Tao;Zhang Xiaochan;Tan Jianxin;Jing Yanwei;Lyu Liangnian(School of Artificial Intelligence,Hebei University of Technology,Tianjin 300401,China;Hebei Jiantou New Energy Co.,Ltd.,Shijiazhuang 050051,China;Goldwind Science&Technology Co.,Ltd.,Beijing 102600,China)

机构地区：[1]河北工业大学人工智能与数据科学学院,天津300401 [2]河北建投新能源有限公司,石家庄050051 [3]金风科技股份有限公司,北京102600

出　　处：《太阳能学报》2024年第11期73-83,共11页Acta Energiae Solaris Sinica

基　　金：河北省科技支撑计划(E2024202051,20314501D,F2021202022)。

摘　　要：为改变“源随荷动”的传统运作模式并增加储能,实现能源网、负荷、储能等各环节协调互动,建立电热氢耦合综合能源系统(ETHC-IES)优化调度,其中应用氢储能实现安全稳定运行的“源-网-荷-储”的新型综合能源系统成为目前的研究热点。以降低综合能源系统运行成本并减少弃风弃光为目标,将ETHC-IES优化调度问题转换为马尔可夫决策过程(MDP),提出应用基于连续动作的近端策略优化算法(PPO)的综合能源系统优化调度方法。首先建立电热氢储能各部分的数学模型,综合考虑功率平衡,安全状态等约束条件,然后采用PPO算法对模型进行求解,以提高经济性和减少弃风弃光为优化目标,重新设计深度强化学习模型的动作空间、状态空间、奖励函数等,智能体通过训练学习实现ETHC-IES的动态调度优化决策。最后,通过仿真验证所提出模型和优化方法的有效性和优越性。In order to change the traditional operation mode of“source with load”and increase energy storage,and to realize the coordination and interaction of power grid,load,energy storage and other links,this paper establishes an electric-thermal-hydrogen coupled integrated energy system(ETHC-IES)for optimal scheduling.The use of hydrogen energy storage to realize the safe and stable operation of a new type of integrated energy system“source network storage”has become a current research hotspot.The aim of this paper is to reduce the operation cost of IES and the waste of wind and light.The ETHC-IES optimal scheduling problem is transformed into a Markov decision process(MDP),and an optimal scheduling method based on continuous action proximal policy optimization(PPO)is proposed for the integrated energy system.Firstly,a mathematical model of each part of the electric hydrogen storage system is established.Then,the model is solved using a deep learning proximal policy optimization algorithm with the optimization objectives of economy and reduction of wind and light waste.The action space,state space,and reward function of the deep reinforcement learning model are set up.Intelligent agents are trained and learned to achieve dynamic scheduling optimization decisions for ETHC-IES.Finally,the effectiveness and applicability of the proposed model and method are verified by simulation.

关 键 词：强化学习 储能 可再生能源 近端策略优化 ETHC-IES 

分 类 号：TK519[动力工程及工程热物理—热能工程]