Open AccessEconomical operation strategy of an integrated energy system with wind power and power to gas technology – a DRL‐based approach
Author(s) -
Zhang Bin,
Hu Weihao,
Cao Di,
Huang Qi,
Chen Zhe,
Blaabjerg Frede
Publication year - 2020
Publication title -
iet renewable power generation
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.005
H-Index - 76
eISSN - 1752-1424
pISSN - 1752-1416
DOI - 10.1049/iet-rpg.2020.0370
Subject(s) - flexibility (engineering) , markov decision process , wind power , benchmark (surveying) , randomness , computer science , mathematical optimization , reinforcement learning , electric power system , power (physics) , energy (signal processing) , process (computing) , markov process , control engineering , engineering , artificial intelligence , mathematics , electrical engineering , statistics , physics , geodesy , quantum mechanics , geography , operating system
With the rapid development of artificial intelligence, adopting advanced deep reinforcement learning (DRL) methodologies to solve the optimisation problem in power systems has become more effective. This study proposes a novel energy control method associated with DRL to solve the economical optimisation problems in an integrated energy system with wind power and power‐to‐gas technology. To consider the randomness of wind power and the flexibility of upper‐level energy prices, the economical optimisation problem is formulated as a finite Markov decision‐making process. Cycling decay learning rate deep deterministic policy gradient (CDLR‐DDPG) algorithm is proposed to obtain the optimal operation strategy. A comparison among different benchmark methods is provided to demonstrate the superiority of CDLR‐DDPG algorithm in optimising an economical problem for the considered system.