Soil Classification by Machine Learning Using a Tunnel Boring Machine’s Operating Parameters

Round 1

Reviewer 1 Report

Soil Classification by Machine Learning Using a Tunnel Boring

Referee's Comments to Manuscript ID applsci-1992428 for Applied Sciences (MDPI Journal)

In the submitted paper, the authors present a prediction methodology based on machine learning to find relationships between the tunnel boring machine's operating parameters and measured engineering characteristics of the ground. According to the abstract and conclusions, the authors expect that their results could help to guide excavation in sections of ground that lack prior geotechnical information. The topic is actual; several authors have recently tried to establish a prediction model for geological conditions based on the operation data of tunnel boring machines and published their results. In my opinion, several parts deserve to be extended:

Introduction:

Several current studies related to this topic can be found in the literature. The authors refer to approximately 15 studies in the introduction but discuss only a few of them. In my opinion, the literature review should provide clear motivation for the work and novelty compared to the previous studies. Some other publications and their outcomes can also be mentioned, e.g., [1, 2, 3, 4].

Data set:

Six representative features were selected from the twenty features measured. Did the authors verify the in uence of the number of selected parameters on the results and their accuracy? Does their choice differ from the other studies in the literature review? Could the authors extend the discussion related to this choice?

In Figure 2, correlations between six quantities are plotted. The title of the image should include the number-feature assignment. Is the order the same as mentioned in the text? Please add a meaning of colours etc. Could the authors show some of the correlation plots for unselected features or comment on them?

The authors mentioned briefly that a Python 3.8 program was used for the data analysis. Is it their own code, open source, etc? In the Modelling methodology, the six employed techniques could be briefly described.

Tables 4 and 5 are incorporated in Section 3.3. Table 5 is referred to in Section 4.1, but the Accuracy defined in Table 5 was presented previously in Figure 4. Please, try to improve the text flow.

Results:

Table 9 shows that the categories based on the N-value of cohesive or non-cohesive soils were reduced to two for each group. Why and is it possible to refine it?

Conclusions:

As mentioned in the conclusion, the results of the study were derived based on data from a single site, and further validation of data from different sites would be needed to assess the accuracy of the prediction. Do the authors consider freely sharing the data or codes used to perform the analyzes presented to support their conclusions?

I recommend major revisions to this paper taking into account the remarks above.

References

[1] P. E. Ayawah, S. Sebbeh-Newton, J. W. Azure, A. G. Kaba, A. Anani, S. Bansah, and H. Zabidi, "A review and case study of artificial intelligence and machine learning methods used for ground condition prediction ahead of tunnel boring machines," Tunnelling and Underground Space Technology, vol. 125, p. 104497, 2022.

[2] H. Xu, J. Zhou, P. G. Asteris, D. Jahed Armaghani, and M. M. Tahir, "Supervised machine learning techniques to the prediction of tunnel boring machine penetration rate," Applied sciences, vol. 9, no. 18, p. 3715, 2019.

[3] H. Yang, K. Song, and J. Zhou, "Automated recognition model of geomechanical information based on operational data of tunneling boring machines," Rock Mechanics and Rock Engineering, vol. 55, no. 3, pp. 1499{1516, 2022.

[4] Q. Zhang, Z. Liu, and J. Tan, "Prediction of geological conditions for a tunnel boring machine using big operational data," Automation in Construction, vol. 100, pp. 73{83, 2019.

Author Response

I am very much thankful to the reviewers for their deep and thorough review. I have revised my present research paper in the light of their useful suggestions and comments. I hope my revision has improved the paper to a level of their satisfaction. Number wise answers to their specific comments/suggestions/queries are summarized in the revised manuscript and attachment.  Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Review report on applsci-1992428：

Soil Classification by Machine Learning Using a Tunnel Boring Machine’s Operating Parameters

The work in its current form does not meet the requirements for publication. My specific comments and suggestions for improvement are as follows:

Abstract

1. The Abstract needs to be rewritten to improve readability.

2. The authors should explain the comparative performance with other models.

3. The authors' methods are inconsistent. For example, light gradient boosting machine (Light GBM) is wrongly written as Light LGM.

Introduction

1. The authors must better clarify the context of this research, especially the significance of the research problem.

2. Contributions need to be emphasized more. It must be made clear what is innovative and how the work overcomes the constraints of previous research.

3. The authors should avoid citing multiple references for one point of view.

4. The authors should introduce the role of each section of the article in detail.

Related Work

1. The authors should add a related work section to introduce the existing methods in detail.

2. The authors should clarify on the distinctions between previous work and the answer offered in this research.

Data Preparation

1. The screening of key features in the dataset needs to be described in detail.

2. As the authors said, " A Python 3.8 program was used for data analysis and machine learning ", the specific process should be explained.

Modeling Methodology and Results

1. The resolution of the figures and diagrams need to be increased. The ones that are currently in the paper are not clear enough and some of them are so small that it has no significance as the information cannot be seen/observed clearly.

2. The authors should indicate the source of literature for each method.

3. The division of training set and test set in the dataset needs to be mentioned.

4. The machine configuration for running the algorithm needs to be mentioned.

5. The authors need to elaborate on the shortcomings of the model and its future progress.

6. The authors should increase the comparison of classification results with the latest methods.

7. The discussion of the results needs to be strengthened.

8. Correct all the grammatical mistakes and language errors throughout the paper.

References

1. The authors need to increase the number of references to support their views, and pay attention to quoting the latest work progress.

The paper may be reconsidered for possible publication once the above-mentioned comments have been addressed in a satisfactory manner.

Author Response

I am very much thankful to the reviewers for their deep and thorough review. I have revised my present research paper in the light of their useful suggestions and comments. I hope my revision has improved the paper to a level of their satisfaction. Number wise answers to their specific comments/suggestions/queries are summarized in the revised manuscript and attachment.  Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The authors have tried to incorporate all my comments. Before publishing the article, I recommend the following:

Improve the quality of the figures, e.g., the numbers and the described units are not clearly visible in Figures 2 and 4. Please, check that you have the column labels correct (F*<50).

In particular, please go through the new parts of the manuscript and correct typos, e.g. "scre torque", missing spaces in quotations, e.g. lines 43, 53, 56, etc., or hard-to-read sentences such as "As mentioned earlier, as the result that the machine data includes ground characteristics [13] Figure 4 shows that each feature has a different distribution depending on the ground characteristics."

Author Response

I have revised my present research paper in the light of your useful suggestions and comments. I hope my revision has improved the paper to a level of your satisfaction. Please see the attachment.

Author Response File: Author Response.docx