A Novel Strategy for Optimal PSO Control Parameters Determination for PV Energy Systems
Round 1
Reviewer 1 Report
In the article authors propose nested PSO for optimization of operation of PV panels. My comments are following:
1) What was the criterion of selectin the benchmark functions. Why only 4 was selected? You should draw them to present their level of complexity.
2) Please unify the unit of PCR because in eq. (3) it is in p.u. and later you use %
3) In Section 3 the most significant parameters should be PCR and computation time. Please highlight this parameters.
4) Equation 4 is not described in text. What is SS?
5) Please explain at the beginning of Section 3 the methodology based on implementation of "CASES". What is definition of single case and what is the purpose of its implementation .
6) Sections 3.4.1; 3.4.2; 3.4.3; 3.4.4 - should be concluded by some comparison by common graph.
7) In theoretical part of this article I can not find the direct comparison between PCR and computation time.
8) Table 6 - please unify the number of significant decimal digits .
9) What was the strategy of selecting the initial position of agents in the swarm.
10) The MPPT operation seems to be not very dynamic. Is it so important to reduce computation time from 5.17 to 2.7 seconds? What is the measurable benefit of shortening calculation time of this problem.
11) In Table 6 the PCR for all strategies is 0% that means that this problem is not a challenge for algorithms. Can you find the case for which convergence will be not 100 % ? It would give more information about the convergence of proposed algorithm for problem of higher complexity.
12) I understand that strategy proposed in this article does optimization of parameters of PSO algorithm. Does strategies described as "S1" to "S10" do optimization of this parameters as well, or they keep them constant?
Author Response
Response to Reviewer Comments
In the article authors propose nested PSO for optimization of operation of PV panels. My comments are following:
1) What was the criterion of selectin the benchmark functions. Why only 4 was selected? You should draw them to present their level of complexity.
Author’s Response: Thanks a lot for this objective comment, Regarding the choosing of the benchmark function, actually the main objective for the paper is to show how to determine the optimal PSO control parameter, and the selection of benchmark functions was based on choosing the most famous ones.
Regarding the choosing only four functions: As you know, it is lengthy papers due to the need for presenting the details of the new strategy itself “NESTPSO” and in case I increased more functions it will make the paper very lengthy and as you can see each objective function needs one table (as tables 1 to 4), moreover, it will need two figures (as the figures 5 to 12), moreover it will need one more figure (as figures from 13-15), moreover it will add 2 columns to Table 5 which will make it more complex and hard for readers to follow. For these reasons, I could not increase the function because simply If I used 8 benchmark functions instead of four I will need more 4 tables, 12 figures, 8 more columns to Table 5, and more discussions about their results which will make the paper in very very lengthy where its number of words in this current status is 12,000 words which is counted as very lengthy. Also, one of the other reviewers asks for reducing the length of the paper. For me, I will not face any problem because NESTPSO deals with any function as a black box and I do not need to do any change to simulate even 100 benchmark functions. But I think your objective comment can be discussed in new research for evaluating these new programs with several functions where I do not have to discuss the program itself as shown in this paper but I can refer to this paper.
Regarding the drawing of these benchmark functions, as you know, they are very famous functions and their 3-D draw in a huge number of papers and for this reason, I did not include them in my previous version, meanwhile, as a respect to your comment, I added a 3-D draw for these functions (Figure-5 in the new version).
2) Please unify the unit of PCR because in eq. (3) it is in p.u. and later you use %.
Author’s Response: Corrected, thanks a lot.
3) In Section 3 the most significant parameters should be PCR and computation time. Please highlight these parameters.
Author’s Response: These parameters are highlighted as you recommend.
4) Equation 4 is not described in text. What is SS?
Author’s Response: I added a Discerption for it in the text in “red”, SS is swarm-size and I defined it in many places in the text and it is shown in a list of symbols and list of abbreviations tables.
5) Please explain at the beginning of Section 3 the methodology based on implementation of "CASES". What is definition of single case and what is the purpose of its implementation.
Author’s Response: I defined these cases inside the text as a list of cases (in red) and these cases are shown clearly in figures 5 to 12, and they are listed and summarized in Table 5. And as you recommend I add points to represent these cases together to make them clearer for readers.
6) Sections 3.4.1; 3.4.2; 3.4.3; 3.4.4 - should be concluded by some comparison by common graph.
Author’s Response: Thanks a lot for this comment, sections 3.5, 3.5.1, and 3.5.2 are used to compare these sections, and Table 5 is used to summarize all the results obtained from the sections you pointed out.
7) In theoretical part of this article I cannot find the direct comparison between PCR and computation time.
Author’s Response: As has been shown in the paper, the main objective of this study is to use a novel study that can reduce the convergence time and the PCR to the lowest value possible. The value of Nss is equal to the number that the NESTPSO hits the objective function to determine the fitness value and we compared these two important issues in the graph from Fig. 5 to 12 in the previous version of this paper and shown in Table 5.
8) Table 6 - please unify the number of significant decimal digits.
Author’s Response: Thanks a lot for this comment, I unified them in all columns of this table.
9) What was the strategy of selecting the initial position of agents in the swarm.
Author’s Response: As you know, we have two nested PSO loops and their initializations were random values within specified limits recommended in the literature, except for the MPPT of PV system we used eq. (14) as has been recommended in many literatures to improve the conversion time and the PCR%.
10) The MPPT operation seems to be not very dynamic. Is it so important to reduce computation time from 5.17 to 2.7 seconds? What is the measurable benefit of shortening calculation time of this problem.
Author’s Response: Thanks a lot for this question, as you know, partial shading can be caused by moving clouds, some trees shadowing which may be very dynamic especially in unstable weather conditions, or in some moveable solar cars, these partial shading dynamics are having very fast change. Moreover, in the real market of the MPPT devices, the convergence time issue is crucial and the quality of the MPPT is assed based on this parameter too. So, convergence time is a very important issue in the MPPT of the PV systems. And as I show in the paper, this MPPT is supposed to be used online (online means: The optimization should be performed while the system is working) which means that we need to get the optimal duty ratio in the shortest time to improve the stability of the PV system quickly. So reducing the time from 5.17 to 2.7 means you reduced the conversion time to half of its previous value which is a very important achievement.
11) In Table 6 the PCR for all strategies is 0% that means that this problem is not a challenge for algorithms. Can you find the case for which convergence will be not 100 % ? It would give more information about the convergence of proposed algorithm for problem of higher complexity.
Author’s Response: First of all, the application showing the PCR values greater than zero as shown in the benchmark functions shown in tables 1 to 5 and figures 5 to 12 and how it works with NESTPSO. Regarding the MPPT is a one-dimensional optimization problem and it is much simple than the benchmark functions shown above, moreover, the initialization of particles based on equation (14) improved the PCR to 0% with many strategies as shown in Table (5). So, the only requirement from the NESTPSO in the MPPT application is to get the optimal parameters and swarm size for the lowest convergence time which has been shown clearly in the results shown in Table (5).
12) I understand that strategy proposed in this article does optimization of parameters of PSO algorithm. Does strategies described as "S1" to "S10" do optimization of this parameters as well, or they keep them constant?
Author’s Response: Thanks a lot for this objective comment, as you know most control parameters of the PSO or any other swarm optimization techniques are determined from tuning criterion or try and error strategy and no formula can determine the optimal values of these parameters that can work with all applications and this was the main motive for me to introduce a new strategy that can determine the optimal control parameters of the PSO or any other optimization techniques which have been achieved by NESTPSO. Regarding the techniques from S1 to S10, they are different control parameters from different researches they introduced them based on tuning criterion or based on try and error and we used them to compare their results with the results from NESTPSO, where these results showed that the NESTPSO achieved better results than these strategies in all benchmark functions or with the MPPT of the PV system.
Finally, the author is would like to thank you for your objective comments that really improved the quality of the manuscript.
Kind Regard,
Author
Reviewer 2 Report
Review Report on Manuscript sustainability-1063600:
It is true that the paper is well written, but there is a lot of papers already published with the same contribution.
Therefore, the author must clearly show his contribution to the literature.
Moreover, I am unable to understand figures 19, 20 and 21, they are correct?
Author Response
Author’s Responses To Reviewer 2
It is true that the paper is well written, but there is a lot of papers already published with the same contribution. Therefore, the author must clearly show his contribution to the literature.
Author’s Response: Thanks a lot for your support in this comment in the written quality of the paper, and thanks a lot for the rest of the comment too, and I really respect your opinion and thought. But, as I said and insist in the abstract, introduction, conclusions, and many parts of the paper that, this paper is introducing a novel strategy that never shown before or any similar work to it in the literature to determine optimal control parameters for the PSO and any other optimization strategy to reduce the convergence time, failure rate. This novel strategy “NESTPSO” showed a great reduction of these issues compared to 10 states of the art strategies that used control parameters obtained from tuning parameters and try and error estimations. The clear, great contributions, and the superior results obtained from this novel strategy “NESTPSO” compared to 10 states of the art strategies proved its superiority and importance that for sure will help researchers, designers, and It for sure will add a considerable value and contribution to the scientific community.
Moreover, I am unable to understand figures 19, 20 and 21, they are correct?
Author’s Response: Thanks a lot for your comment, these figures are shown before in many studies that show the time variation of the power (the curve in the left of these figures) and duty ratio (the curve at the bottom of these figures) connected to the power-duty curve of the PV array (the curve in the top-right of these figures). Each point on the power-duty curve should have reflections in the two figures otherwise the simulations are not correct. This figure is discussed in more details in the research shown below if you need further details about it:
Eltamaly, Ali M., et al. "Simulation and experimental validation of fast adaptive particle swarm optimization strategy for photovoltaic global peak tracker under dynamic partial shading." Renewable and Sustainable Energy Reviews 124 (2020): 109719.
Finally, the author is would like to thank you for your objective comments that really improved the quality of the manuscript.
Kind Regard,
Author
Reviewer 3 Report
The article covers the issue of the optimization of PSO control parameters in application to the photovoltaic systems maximal power point tracking. I find the whole article interesting, however the text is quite long and detailed. The disproportion between the part of the article related to the development and testing of NESTPSO and the part concerning the PV application is visible (15 p. vs 4 p.) and might be moderated by several additional information about the application, especially in the context of possible profits in precisely tracked PV systems, which would fit into the journal main characteristic.
Several details:
- "The PV system is shown in Figure 17 is showing PV" - (grammar)
- Table 6. The performance evaluation of the PSO with state-of-the-art strategies and the NESTPSO: "column: % reduction compared to NESTPSO...
"% reduction compared to NESTPSO" suggests that the given strategy reduces the time, which is not true.
Please try to change the headline of this column in order to show the benefit of NESTPSO (i.e. "% extension" or "% reduction by NESTPSO")
- Figures 19-21 - is it possible to obtain the predicted energy benefit (kWh) from this simulation?
- (602)The private PSO parameter cl is sometimes going to a negative value which is not sown before in the literature. - shown*?
- Please format the "REFERENCES" according to the journal guidelines, i.e. using ZOTERO or other applications.
Author Response
Author’s Responses To Reviewer 3
The article covers the issue of the optimization of PSO control parameters in application to the photovoltaic systems maximal power point tracking. I find the whole article interesting, however the text is quite long and detailed. The disproportion between the part of the article related to the development and testing of NESTPSO and the part concerning the PV application is visible (15 p. vs 4 p.) and might be moderated by several additional information about the application, especially in the context of possible profits in precisely tracked PV systems, which would fit into the journal main characteristic.
Author’s Response: Thanks a lot for this objective comment, actually, the main object of this paper is to determine the optimal control parameters of the PSO optimization technique and it should be discussed and explained in details and applied to several benchmark functions and for this reason we used 4 benchmark functions for comparison, where any additional function will need extra 4 figures and one more table which will increase the “NESTPSO” discerption part and for this reason, it is not easy to reduce this part. Moreover, one of the other reviewers recommend to use more benchmark functions which will increase this part too. Regarding the MPPT of the PV system it is introduced here in this paper to prove the right results from “NESTPSO” as any benchmark functions shown above and for this reason it takes 4 pages. Adding more parts in the PV part will make the paper very very lengthy which is not favorite as you recommend in the first part of your comment.
Several details:
- "The PV system is shown in Figure 17 is showing PV" - (grammar)
Author’s Response: Corrected
- Table 6. The performance evaluation of the PSO with state-of-the-art strategies and the NESTPSO: "column: % reduction compared to NESTPSO..."% reduction compared to NESTPSO" suggests that the given strategy reduces the time, which is not true.
Author’s Response: Corrected.
Please try to change the headline of this column in order to show the benefit of NESTPSO (i.e. "% extension" or "% reduction by NESTPSO")
Author’s Response: Thanks a lot for this comment, I changed it.
- (602)The private PSO parameter cl is sometimes going to a negative value which is not sown before in the literature. - shown*?
Author’s Response: This is one of the benefits obtained from NESTPSO where the negative values of cl improved the convergence time and PCR which was not shown before in any literature.
- Please format the "REFERENCES" according to the journal guidelines, i.e. using ZOTERO or other applications.
Author’s Response: Modified, thanks.
Finally, the author is would like to thank you for your objective comments that really improved the quality of the manuscript.
Kind Regard,
Author
Round 2
Reviewer 1 Report
The benefits from the reduction of computation time could be described more significantly.
Article can be accepted in present form.
Reviewer 2 Report
The revisions are ok.
