Review Reports - TC-SOM Driven Cluster Partitioning Enables Hierarchical Bi-Level Peak-Shaving for Distributed PV Systems

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The paper offers an innovative approach to flexible peak-shaving in power systems with high levels of distributed PV penetration by integrating a neural network, namely the Temporal Coupling Self-Organizing Map, into a bi-level control model. In this regard, the TC-SOM algorithm performs hierarchical PV cluster partition based on electrical coupling, power balance, and temporal characteristics, while the bi-level model coordinates peak-shaving from node demand to the cluster and individual PV levels. For this methodology, the validity is verified using the IEEE-33 node test system, which demonstrates better smoothing of loads, a reduced peak-valley difference, and increased flexibility compared to conventional approaches. According to the authors, this strategy further contributes to more efficient and stable grid operation with high PV integration. The article should undergo a major revision based on the following comments:
1. The topic of distributed PV-based peak-shaving is timely and relevant. The idea of employing TC-SOM with a bi-level control structure is an interesting concept in itself.
2. The paper is well written and clearly organized, but the technical contribution is not strong enough to warrant publication in its current form.
3. There are no figures or graphical results in the manuscript, which is a serious omission for a work that relies so heavily on algorithmic modeling and simulation. The claimed improvements, without the visual evidence of clustering results, load profiles, or comparison plots, cannot be independently verified.
4. While the TC-SOM model is extensively presented here, it nevertheless seems like an incremental adjustment to the traditional SOM framework rather than an important methodological novelty. This paper largely rewraps the old ideas into a new combination without offering new theory or algorithms.
5. The section on simulation and validation is weak. Descriptions of all the performance comparisons are given in words only, without quantitative tables or graphical metrics. It is not clear how results like reduction rates or efficiency enhancements were calculated.
6. The test case used is the IEEE-33 node system, but details on parameters used in simulations, capacity of the PV, datasets used, or environmental assumptions are not provided. The lack of these specifications undermines reproducibility.
7. The mathematical presentation is generally clear, but variables are either undefined or defined inconsistently between equations, which makes following derivations quite hard. A list of symbols and their definitions should be supplied.
8. The manuscript contains a large number of equations; however, it does not indicate how they are implemented and numerically validated. There is no discussion on algorithm convergence behavior, computational complexity, or sensitivity to the choice of parameters.
9. Comparison with other methods, like K-means, is only briefly mentioned; no numerical benchmarks or figures are provided. The improvement in percentages appears arbitrary without data visualization to support such claims.
10. The bi-level model structure is conceptually sound but already a usual approach in hierarchical control and optimization studies related to distributed energy systems; the novelty therefore lies more in the application context than in technical development.
11. The writing is generally fluent. However, there were several sections that were wordy and could be reduced to make the text clearer. Long sentences and repeated descriptions of the same concept makes reading difficult.
12. The discussion only provides descriptive statements of improvement, such as reduction percentages, with no deeper physical explanation or analysis of why the proposed model performs better.
13. This absence of figures is all the more disturbing because the paper mentions several flowcharts, network diagrams, and output curves that are never presented. These omissions preclude any judgment of correctness or realism by the reader.
14. The reference list is reasonably up to date, covering recent publications, but the citations are dominated by Chinese and internal institutional sources. Broader international references on hierarchical control or advanced SOM variants would strengthen the background.
15. The structure of the paper follows the general format of an engineering study, but the validation does not meet the standard of this journal for reproducible scientific evidence. Quantitative evaluation, figures, and better demonstration of novelty are required.
16. The paper, in its present form, is more of a technical report on an algorithmic framework rather than a scientific contribution with verified results.
17. This could be a good paper with visual results, detailed experimental parameters, and a stronger case for originality from the authors.
18. However, in the absence of figures, with limited validation, and innovation not clearly demonstrated at present, the manuscript is not ready for publication in its current form.

Comments on the Quality of English Language

The authors should check the whole manuscript for typos and grammatical errors.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

- The introduction section adds an evaluation of other optimization methods and describes the reasons for choosing the IEEE-33 node system for the study.
- The author should rename section 2 to Methodology.
- Additional mathematical formulas cite references (if any).
- Add units for operators in the formulas.
- Figures 8, 9, 10, 11, 13, 14, 15, 16, 17, and 18 are resized for easier viewing.
- Section 4.1. adds technical specifications of PV projects.
- The research results should have discussion comparing and evaluating with research works on the same topic to clarify the feasibility of the proposed method.
- Check the information of reference document No. 23.

Comments on the Quality of English Language

English could be improved.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

On the claimed gap regarding PV peak-shaving capability:
The authors state that previous studies mainly focused on coordinated peak-shaving strategies of distributed PV with other equipment, neglecting the intrinsic peak-shaving capability of PV itself. However, PV systems inherently lack active ramping or peak-shaving capability, as their output is determined by solar irradiance and can only be curtailed (not freely adjusted like gas turbines or ESSs). Therefore, this limitation cannot be “solved” by the proposed method. It is suggested that the authors revise the problem statement to avoid misleading claims and clarify what aspect of PV flexibility their model actually improves (e.g., coordinated control, aggregation optimization, or curtailment management).
On the clustering approach:
The idea of dividing PV units into clusters is potentially interesting, yet its motivation and necessity are not well justified. Modern distribution systems equipped with edge computing and advanced communication infrastructures can already support individual PV-level control. Therefore, the benefit of clustering needs to be explicitly demonstrated.
- If the clustering is introduced to mitigate dimensionality or computational scalability, the experiments should be performed on large-scale PV networks, rather than the simple IEEE 33-bus system.
- For such small-scale systems, existing methods (without clustering) could achieve similar results, as demonstrated in the following reference: DOI:10.1109/TITS.2023.3314571
On contribution clarification and introduction restructuring:
The Introduction should be rewritten to include a more comprehensive literature review and to clearly differentiate the proposed approach from existing peak-shaving or PV coordination methods. The novelty and practical significance of the TC-SOM-based clustering and bi-level framework should be made explicit—what specific problem does it solve that prior works could not?

Morover, Figures: The legends and text in figures are too small and unclear. Please enlarge the labels for readability.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

The sentence in the abstract "This hierarchical structure inherently embodies symmetry in response logic-realizing balanced interaction..." is a bit wordy.
Some of the sentences in the abstract are a bit long, especially in the introduction and modeling section (lines 9-18). Shortening these sentences could have increased readability faster.
Lines 575-576 state that "Since the disturbance occurs in Cluster 3, the photovoltaics within this cluster operate at full output at all times..." This section correctly reflects the role of the affected cluster. However, in Table 3 (Scheduling Distance), Cluster 3 has a Scheduling Distance of zero (0) (line 569). This zero distance implicitly indicates that this cluster is the closest or affected node. The cause-and-effect relationship between "zero distance" and "working at full capacity" should be explained more clearly in the text (why does zero distance mean no interference with peak shaving?).
The TC-SOM clustering results (dividing 11 nodes into 4 clusters) are repeated in two different places (lines 507-508 and lines 541-542) with almost the same details. This repetition could be reduced.
In the final comparison (lines 617 to 627), the superiority of the proposed model over "Traditional Peak Shaving Methods" is discussed. It is not clear which models are exactly meant by these traditional methods. For a scientific paper, it is necessary to specifically mention the baseline models in order to have a valid scientific comparison.
In paragraph (2) of line 658, the rate of decrease of the peak-valley difference is mentioned as 13.62%. This number is a bit confusing compared to the figures presented in the results section (which referred to a rate of decrease of 21.64% for the imbalance index and 11.72% for the peak-valley in the time window 05:00-18:00 and 21% compared to the traditional method). I suggest you make sure that this number (13.62%) exactly refers to which scenario and which criterion, or it is better to use more prominent and uniform figures (for example, 21% obtained compared to the traditional method) so that the final message is stronger.
Paragraph (3) (lines 662-667) is largely a repetition of the same importance and value of engineering as stated at the end of the abstract (lines 22-25) and the initial conclusion paragraph (lines 644-647). I suggest reducing paragraph (3) to a shorter sentence about "providing a roadmap for future work" or "long-term engineering value" to avoid mere repetition.
All figures from 14 to 17 are not clear. And figures 10 and 11 and re-export. The figures are not acceptable at all at the moment. I cannot check
For figure 8, first use colors that ensure the text is in black. The figure is not acceptable at all.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

All comments have done.

Comments on the Quality of English Language

The authors should check the whole manuscript for typos and grammatical errors.

Author Response

Comment 1: All comments have done.

Author response: We sincerely appreciate the valuable and constructive comments you have provided regarding our manuscript. Your profound insights have played a crucial role in enhancing the completeness, depth, and academic rigor of the paper. By adopting your suggestions, we have supplemented a comprehensive evaluation and comparison of representative optimization methods in the case study section, and added relevant technical parameters. This has significantly enhanced the effectiveness and universality of the strategies we proposed. Your meticulous review and thoughtful feedback have helped us refine the research framework and improve the overall quality, making the manuscript more rigorous and persuasive. We are extremely grateful for the time and effort you have invested in our paper. Your contribution is crucial to the successful revision of this study.

Comment 2: The authors should check the whole manuscript for typos and grammatical errors.

Author response: Thank you very much for your meticulous and valuable comment. In response to your suggestion, we have conducted a comprehensive, sentence-by-sentence review of the entire paper: we corrected minor typos, grammatical inconsistencies, and unified technical term spellings; adjusted improper punctuation to improve readability; and verified the consistency of formula notations and figure captions to eliminate ambiguity. These revisions not only address structural optimization but also ensure the overall accuracy and clarity of technical expressions, making the manuscript more polished and convincing. We sincerely appreciate your careful review, which has played a crucial role in refining the quality of this work.

Author action: We standardized the spelling of technical terms, such as converting "peak reduction" to "peak elimination", "K-means" to "K-mean", and "two-layer model" to "two-layer model". We also corrected spelling errors like "lo-cations" being merged into "locations" and "reductio" being corrected to "reduction". We also fixed grammatical inconsistencies throughout the manuscript and added clear units for all mathematical formulas, such as the unit of the sensitivity matrix S being kV/kW, the unit of system load being kW, and the unit of photovoltaic self-consumption output being kW/h, to eliminate ambiguity, standardize punctuation, remove unnecessary spaces in "first-layer" and "inter- and intra-cluster", add commas in lists, correct incorrect table references, update "Table 3" to "Table 5", "Tables 4 and 5" to "Tables 7 and 8", optimized the reference format, added the complete DOI prefix https://doi.org/, and supplemented the publication month of the references.

Reviewer 2 Report

Comments and Suggestions for Authors

The authors revised the paper according to comments of reviewer.

Comments on the Quality of English Language

English could be improved.

Author Response

Comment 1: The authors revised the paper according to comments of reviewer.

Author response: We are deeply grateful for the incisive and constructive feedback you furnished on our manuscript. Your expert observations were pivotal in sharpening the paper’s scope, depth, and scholarly rigor. Acting on your recommendations, we have expanded the case-study segment with a systematic benchmarking of state-of-the-art optimization techniques and appended all pertinent technical specifications, markedly strengthening the practicality and generalizability of our proposed strategies. Your painstaking review and discerning suggestions have allowed us to recast the research framework and elevate the overall quality, rendering the work more compelling and robust. Thank you for the considerable time and intellect you devoted to this revision; your contribution has been instrumental to its successful refinement.