Long-Term Structural State Trend Forecasting Based on an FFT–Informer Model
Round 1
Reviewer 1 Report
· - The concept of structural state forecasting should be well defined in the Introduction part. What are the key parameters to be assessed? How can they be associated with the structural-state conditions?
· - The statement between the lines 40- 43 cannot be understood well. Please revise this statement for better understanding of the reader.
· - What is ProbSparse? Please shortly describe in the text.
· -At the first stage of the implementation, do you use overlapping segments while windowing the data?
· -Please check the accuracy of Eq. (1).
· - Please provide a more specific definition for Am in Eq. (2). What kind of feature does it actually represent?
· - How to fit the data by Eq. (3), please elaborate.
· - Please elaborate more the problem of Multi-head ProbSparse self-attention.
· - In Eq. (4), what is softmax? Does it represent a specific programming function? Please elaborate?
· - Please define the dimension of the vectors qi , ki , vi.
· - Please make a more specific description about the input parameters.
· - The proposed method is verified with experimental Benchmark studies only? I recommend to include at least one numerical example to show its performance in the presence of noisy data? How it performs in case of increasing noise levels?
Author Response
Thanks for pointing out these issues.
Please see the attachment
Author Response File: 
Author Response.docx
Reviewer 2 Report
Comments for author File: 
Comments.pdf
Author Response
Thanks for pointing out these issues.
Please see the attachment
Author Response File: Author Response.docx
Reviewer 3 Report
1. Please remove the "we" from your writing and use passive voice
2. In general, there are long sentences in your abstract and Introduction. Please minimize the length and be more focused.
Please identify each abbreviation during writing, for example RNN, AND LSTF!!
Author Response
Thanks for pointing out these issues.
Please see the attachment
Author Response File: Author Response.docx
Reviewer 4 Report
Title: Long-term Structural State Trend Forecasting Based on FFT-Informer Model
The article demonstrates the application of a model using fast Fourier transforms and Informers to the forecasting of structural health information. A detailed description of the proposed machine learning model is offered. Subsequently, said model is applied to two benchmark data sets. The paper is certainly interesting and it demonstrates a promising approach to obtain realistic structural health forecasts. The manuscript could, however, benefit from some additional clarifications regarding the implemented model as well as from a grammar revision. For these reasons, the article is recommended for publication after a minor revision.
Some additional comments below:
- It is acknowledged that the authors present a detailed description of their proposed model. However, it may be difficult to understand for the colleagues in the structural engineering community who are not experts in machine learning. For this reason, it is suggested that the authors rework parts of section 2 with said colleagues in mind. Perhaps it would be useful to reference figure 1 throughout he section to make clearer the flow of information. Additional discussion regarding the motivation for each of the components of the model may be useful as well.
- More discussion is necessary regarding the model’s parameters, their significance and the sensibility of the results to changes in one or more parameters.
- Additional discussion should be offered regarding the applicability of the forecasts produced by the model in the context of the selected data sets.
- The text and tickers are too small in all figures. Those should be made larger. The units of measurement should be explicit in the x and y axes of every figure.
- The text requires a review of the use of the English language. While there are no glaring grammar mistakes, there are some odd sentence structures. Additionally, the authors commonly use the singular form while the plural would be more appropriate (e.g., in line 50 it should read: Transformer models with encoder-decoder systems). Moreover the use of definite and indefinite articles must be revised.
Author Response
Thanks for pointing out these issues.
Please see the attachment
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
The authors have addressed all the concerns by the reviewer. The manuscript can be accepted as is.
Author Response
Dear Reviewer，
Thank you for your review. In this manuscript, we also revised some contents to reduce the repetition rate and make the description more clearer, which is marked red in the text.
Please accept our warmest regards.
Sincerely yours,
Jihao Ma
Reviewer 3 Report
1. Author did not remove the word "we". Passive voice must be used!!!
2. Equation 4, please complete the missing part. You have used (??) rather than pointing to the equation number.
3. all equations must have reference
Author Response
Dear reviewer,
Thank you for your review. Please see the attachment.
Please accept our warmest regards.
Sincerely yours,
Jihao Ma
Author Response File: Author Response.docx
