Research on Power Cable Intrusion Identification Using a GRT-Transformer-Based Distributed Acoustic Sensing (DAS) System

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

-The abstract is informative but overly technical. Simplify the language to make it more accessible to a broader audience. Clearly state the novelty of the GRT-Transformer model upfront.

- Provide more details about the dataset, such as the distribution of intrusion events and any preprocessing steps applied.

- The comparison of denoising algorithms (Fig. 9, Fig. 10) is useful, but the results could be summarized in a table for clarity.

- The confusion matrices (Fig. 13, Fig. 14) are informative, but the text should explicitly discuss the misclassifications and their implications.

- The comparison with other models (Table IV, Fig. 15) is thorough, but the discussion should delve into why the GRT-Transformer outperforms others (e.g., specific architectural advantages).

- Provide full details of the GRT-Transformer architecture, including hyperparameters and training protocols.

The manuscript presents a novel and an innovative approach to intrusion detection in DAS systems. However, it requires major revisions to improve clarity.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The paper addresses a problem that I believe is of great interest: intrusion detection systems for power cables. Unfortunately, I could not find much (if any) evidence that the proposed methodology to improve the high false alarm rate other than the report in the paper.

Here are some major points the authors must address to improve the quality of the paper:

• How was the proposed methodology compared with other methods? Apparently the authors collected their own data, which is not provided. If data cannot be provided (it would be desirable!), then please explain how to used the other methods on your own data.
• Provide a better description of your methodology (code would be desirable) and how it performs on other datasets. Is your method faring better, in general?

Here are some more remarks to improve the clarity of the paper. I try my best to indicate locations, but the paper pagination is completely messed up.

• Abstract: why Gramian Angular Field is abbreviated as GASF? Did you miss a word, that appear someplace later in the presentation?
• Keywords: please use "phase" instead of Phi symbol; keywords are typically used for searches, a symbol would be of no use.
• Introduction, at the very end: no mentioning of the VMD-SWTTV algorithm but a few pages later it is presented as a contribution of the paper. Then your summary of the paper does not match the actual structure of the paper. 
• Page 6 (then throughout the paper): tables are reference using Roman numbers (Table I, Table II, etc.), but they are labeled using Arabic numbers (Table 1, Table 2, etc.).
• Page 7: Algorithm 1 presentation must be improved (use indentation, equations in their own line, etc). Whose algorithm flowchart is presented in Figure 4? The GRT-Transformer is presented as Algorithm 1, then the discussion seems to be focused on VMD-SWTTV. Some clarity is needed.
• Section 3: There are many details (formulas) for normalization, polar coordinates transformations, BiLSTM, etc. They are from literature, and I don't see their relevance. In subsection 3.5, where the proposed GRT architecture is proposed, it would be great to emphasize the novelties. What's specific about the proposed architecture, other than a network constructed in a very specific way from standard modules?
• Page 8 (and more): figures are too big to either fit in a page or appear clearly. What use of so much details if they cannot be seen?
• Page 8, etc: equations should be better align (to the left)
• After page 12: page numbering starts over?
• Section 4.2: There are a lot of denoising algorithms, citations would help. How does Figure 9 prove the superiority of one algorithm vs. others? Other than claiming that, you should probably justify using some metrics. What is "k" in equation (24)? Figure 10 does a better job justifying the claims, but how statistical relevant are your 10 samples?
• Can you please separate Figures 13 and 14? Not only they appear together, but it looks like Figure 14 is between parts (a) and (b) of Figure 13.
• I find a little bit unusual that your training set accuracy is lower than your testing set accuracy. Why would that happen?

Comments on the Quality of English Language

Language is fine, in general (no typos). But the paper needs a careful revision as there are quite a number of repetitions (for instance, your Conclusion is nothing much than abstract and some parts of introduction, which a few changes here an there), phrases that need to be reformulated to make sense (see your very first phrase in the Abstract), orphans (like around Algorithm 1), etc.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The authors proposed a GRT-Transformer-Based Distributed Acoustic Sensing System to identify power cable intrusion. The paper requires significant improvement, especially in terms of writing and organizing aspects. Please address the following comments for further evaluation.

1. The authors stated in the introduction that the multi-modality concept is underexplored. However, what is the meaning of investigating the effect of multi-modality if a problem has already been accurately solved with a single modality? As it is indicated in the Introduction that it is the literature lacks such exploration, please provide an extensive discussion on that matter based on the analysis results.

2-What is the basis of selecting the methods for comparison regarding Table 4?

3-How suitable is the model for a real-world application? Please discuss it in terms of computational cost and performance together.

4-Figure 3 is a bit hard to read, please change the color selection as well as improve the quality by tuning the font size and other issues.
Check the grammatical errors and typos within the manuscript.

5- The organization and scientific writing should be significantly improved.

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

I have no further queries.

Reviewer 2 Report

Comments and Suggestions for Authors

I find the presentation of the second draft of the article greatly improved. The authors satisfactory addressed/answered all questions and comments.

The figures are much better presented and easy to understand. Unfortunately, some are way out of the page range. Some can be split into more rows, while others may be displayed in landscape layout?

Please use a space between content and the citation brackets, throughout the paper.