Agent-Based Energy Sharing Mechanism Using Deep Deterministic Policy Gradient Algorithm
Round 1
Reviewer 1 Report
The article presents the energy sharing mechanism facilitating the consumption of local energy, which is formulated as a leader-following model based on the theory of Stackelberg games. In order to solve it, a deep deterministic policy gradient was used with the free approach to the model based on deep reinforcement learning, without the need to have any information about the lower model. The process of solving the nash equilibrium problem was presented (based on the deep deterministic policy gradient algorithm and the effectiveness of the proposed approach was demonstrated)
The article deals with an interesting, not widespread approach to the subject, which is not devoid of research significance and could be applied utilitarian.
My comments to the article:
- Line 17, 25, 88 - no explanation of the abbreviations used, e.g. ES, RES, MDP, which may cause difficulties in reading the text
- Line 301- Figure 7 - on the ordinate axis is the description of iteration, while in the description line 296 is the time interval.
- Line 303 - Figure 8 - I do not understand what it is supposed to present, the description in the point is unclear.
- Line 332 - there is a description to Figure 11 which is only placed in line 356
- Line 334-350 - the authors prove that the deep Q network (DQN) method is ineffective, while in the further part (fig 12) they refer to it.
- Conclusion - the section can be improved by focusing on a broader description of the obtained effects and their possible consequences for prosumers, and on the importance of the applied solution in the field of scientific research.
Comments for author File: 
Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
introduction should be improved, recent work on energy management techniques, agent-based techniques benefits, technique, application on energy management.
English format should be reviewed, for example leave space before "(".
avoid bullets and convert into tables and text
please explain if the proposed algorithm in figure 3 can be possible without agents.
change title of section 5, from numerical results to "results"
the conclusion did not discuss how agents are used, if so, better to change title and focus of the paper.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Reviewer 3 Report
The article presents the agent-based energy sharing mechanism Using Deep deterministic policy gradient algorithm. The problem is formulated as a leader-follower framework based on the Stackelberg game theory. The article is well written and presented timely research however lacking in some aspect. Following suggestion authors can consider revising the submitted manuscript for quality improvement.
- Please modify the sentence ‘Any mismatch between energy supply and demand would increase the cost to both sides.’ since balancing not only affecting the cost but also create technical issues such as stability.
- Typographical errors should be removed e.g. provide space b/w the text and the citation, text and abbreviations, different colored text is also observed,
- The introduction section of the manuscript should be enhanced by adding a paragraph on the latest literature of P2P strategies, algorithms and schemes and then justify the novelty of the proposed method. Following references can be useful if you like. https://www.sciencedirect.com/science/article/pii/S0306261918303398 https://www.nature.com/articles/s41560-017-0075-y https://www.sciencedirect.com/science/article/abs/pii/S0306261919310736 https://ieeexplore.ieee.org/abstract/document/8279516 https://ieeexplore.ieee.org/abstract/document/8274546 https://ieeexplore.ieee.org/abstract/document/8398582
And many more can be found on the google scholar.
- Mathematical expressions and equations seem like images and look blur.`
- The conclusion should be improved by adding the finding of the paper, it should not repeat the abstract. Bullet points can be used.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
accept
