A High-Robust Displacement Prediction Model for Super-High Arch Dams Integrating Wavelet De-Noising and Improved Random Forest

Round 1

Reviewer 1 Report

1. The introduction, algorithm principle, and conclusion are lengthy, and attention should be paid to concise problems;

2. 2.1 The first three paragraphs of the section are irrelevant;

3. The format is chaotic, such as line 276 does not need to be indented,  line 280 text alignment is incorrect, the alignment of the image name in Figure 9 and Figure 10 is inconsistent, and line 688 lacks a stop. There are many similar problems in the full text. It is recommended to check carefully;

4. The abnormal value detected does not participate in the regression. How much can the prediction accuracy of the model be improved? It is suggested to reflect it in the calculation example; 5. The radar chart of the predicted performance index is relatively chaotic, so it is recommended to delete irrelevant lines and re-color;

6. In section 6.3, the model after wavelet denoising is compared with other models without any pretreatment, and it is concluded that the prediction performance of the model after wavelet denoising is better, which is obviously lack of persuasion and credibility;

7. This paper uses a variety of optimization algorithms. Why is the convergence performance of ISSA and GWO only compared in Figure 12?

8. In Section 6.4, three different types of noise are introduced into the model to verify its robustness. How representative are the three types of noise in the dam safety monitoring data? How robust is the model if different types of noise are combined together?

9. Can we further separate the amplitude and compare the interpretation function of the model?

10. There are many problems with grammar and syntax. It is suggested to revise and polish further.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper requires major revisions:

1-Why dis authors use Random Forest for high-robust displacement prediction model rather than SVM, MARS, DT, RT??

2-As an expert in statistical analysis, authors are recommended to checked out the correctness of Eq.(38). I suppose it is incorrect.

3-Various applications of Machine Learning (ML) Models into water engineering problems can enhance quality of literature review:

-A novel multiple-kernel support vector regression algorithm for estimation of water quality parameters

-Riprap incipient motion for overtopping flows with machine learning models

4-What are setting parameters of Random Forest models for the present study?

5-Why did authors apply SSA and GWO algorithms??

6-How is GWO/SSA applied into normal structure of RF??

7-How is Wavalet De-nosing run? What are decomposition levels of Wavelet performance?

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

All modifications meet my requirements

 

Reviewer 2 Report

Accept as is