A Data-Driven Indirect Approach for Predicting the Response of Existing Structures Induced by Adjacent Excavation
Round 1
Reviewer 1 Report
Figures 5 & 7 are looking like buried. Please show the clear figures.
Author Response
Thank you for your suggestions. These two figures have been redrawn to improve their quality.
Reviewer 2 Report
The paper is on prediction the response of existing structures induced by adjacent excavation. Three models for predicting the settlement of building induced by adjacent excavation are provided, namely SVM model, BP model and BP-SVM model. The prediction accuracy of these models is compared and discussed. The method is based on the machine learning.
There are some very related studies in the literature on this subject. These are listed below :
Analysis of Diaphragm Wall Deflection Induced by Excavation Based on Machine Learning, DOI: 10.1155/2021/6664409
Auto machine learning-based modelling and prediction of excavation-induced tunnel displacement, DOI: 10.1016/j.jrmge.2022.03.005
Optimized extreme gradient boosting machine learning for estimating diaphragm wall deflection of 3D deep braced excavation in sand,
DOI: 10.1016/j.istruc.2022.10.027
An improved artificial bee colony-random forest (IABC-RF) model for predicting the tunnel deformation due to an adjacent foundation pit excavation, DOI: 10.1016/j.undsp.2021.11.004
Predicting Excavation-Induced Tunnel Response by Process-Based Modelling, DOI: 10.1155/2020/9048191
Mitigating tunnel-induced damages using deep neural networks,
DOI: 10.1016/j.autcon.2022.104219
The similarities and differences between the current study and the above references should be given and discussed.
Three different models presented here do not give remarkable difference (Figure 9). Why? It should be explained. Then, the meaning of the using different models also should be explained.
"Figure 7" is difficult to understand and the quality of this figure should be improved.
There are some writing errors and the English of the paper should be improved.
Author Response
Q1. There are some very related studies in the literature on this subject. The similarities and differences between the current study and the above references should be given and discussed.
Answers: Thank you for recommending these literatures. We have downloaded and learned these documents and cited them in the revised version. The differences between this article and these documents are described in the revised version.
Q2. Three different models presented here do not give remarkable difference (Figure 9). Why? It should be explained. Then, the meaning of the using different models also should be explained.
Answers: Thank you for your question. Figure 9 shows the prediction results using three machine learning models. Because the same training set and test set are used, the prediction results are not particularly different. The purpose of using three machine learning models for prediction is only to test the influence of machine learning algorithm on prediction accuracy, which will help us determine the direction of follow-up work.
Q3. "Figure 7" is difficult to understand and the quality of this figure should be improved.
Answers: Thank you. We have redrawn this figure.
Q4. There are some writing errors and the English of the paper should be improved.
Answers: Thank you. It is improved as much as we can.
Reviewer 3 Report
Dear Authors,
The manuscript used SVM, BP and BP-SVM machine learning models to predict the settlement of building induced by adjacent excavation. And verified the models with a real excavation project. The manuscript is qualified for publication in applied sciences, however based on the comments attached, I recommend this manuscript after addressing the major revisions.
Thanks!
Comments for author File: 
Comments.pdf
Author Response
Q1. The abstract needs to be revised. The outcomes briefly need to be discussed.
Answers: Thank you. They have been revised.
Q2. Not a very profound literature review was conducted in this introduction. The introduction lacks the recent research works. There are plenty of the research papers that can be discussed in the introduction.
Answers: Thank you for your comments. The relevant references have been supplemented and rearranged in the revised version.
Q3. Line 36: analytical approach, empirical formula, numerical simulation. Please mention the name of different methods in each category with an appropriate reference. For example for empirical methods Peck 1969 method. See the following papers.
Answers: Thank you for your comments and recommending these literatures. It has been supplemented in the revised version.
Q4. The paper lacks the recent research papers on machine learning methods for prediction of the ground response due to the excavation. There are plenty of them.
Answers: Thank you for recommending these literatures. Relevant research status has been supplemented in the revised version.
Q5. Sections 2.21, 2.2.2, again I don’t see the recent literature revie on these methods.
Answers: Thank you for your comments. These two sections only introduce the two traditional mechanical learning methods used in this paper. This study is only a simple use of it, without improving it.
Q6. Line 205 lacks reference.
Answers: Thank you for your comments. It is just a general understanding, without reference.
Q7. Line 208 “The larger the value of ground sub-…” is not clear. Please revise.
Answers: Thank you for your comments. It has been revised.
Q8. Line 228 “Therefore, it can be defined as…” it is not clear what the authors are refereeing to. Please revise.
Answers: Thank you for your comments. It has been revised.
Q9. Line 239 “Deflection of retaining…” lacks reference.
Answers: Thank you for your comments. The treatment method is based on the engineering experience, and there is no reference to discuss this issue.
Q10. Line 247 “The earth pressure on…” lacks reference.
Answers: Thank you for your comments. The treatment method is based on the engineering experience, and there is no reference to discuss this issue.
Q11. Line 264 “Hydraulic head/ground…” lacks reference.
Answers: Thank you for your comments. The treatment method is based on the engineering experience, and there is no reference to discuss this issue.
Q12. Line 266 “The change of hydraulic…” lacks reference.
Answers: Thank you for your comments. The treatment method is based on the engineering experience, and there is no reference to discuss this issue.
Q13. Line 284 “There are two normaliza…” lacks reference.
Answers: Thank you for your comments. The treatment method is based on the treatment method is based on consulting experts, and there is no reference to discuss this issue.
Q14. Line 291 “such as Min-Max normalization…” reference for each method or use a general reference for them all.
Answers: Thank you for your comments. The references have been supplemented in the revised version.
Q15. Line 293 “including premnmx function…” lacks reference.
Answers: Thank you for your comments. The references have been supplemented in the revised version.
Q16. Line 315 “16.12m from the main body of the station” distance from where to where? I would suggest to show the geometries in a schematic figure in addition to figure 7.
Answers: Thank you for your suggestion. It is marked in figure 7.
Round 2
Reviewer 2 Report
Authors made the necessary modifications.
Reviewer 3 Report
Dear Authors,
Thank you for the manuscript. I would recommend this manuscript for publication.
