Service Performance Evaluation of Operating Loess Railway Tunnel Based on Bayesian Network

Round 1

Reviewer 1 Report

I appreciate the chance to read this informative manuscript.

regards 

Comments for author File: Comments.pdf

Author Response

Thank you for your response and suggestions. Here is my reply one by one:

1. Please see manuscript for details;
2. This is a good suggestion. I have revised the conclusion. Please see manuscript for details;
3. Some articles from 2020-2022 were added to the article;
4. Thank you for your affirmation;
5. The MCMC method uses the Monte Carlo integration of the Markov chain. The basic idea is to construct a Markov chain to make its stationary distribution a posteriori distribution of the parameters to be estimated. The posterior distribution samples are generated by this Markov chain, and Monte Carlo integration is performed based on the samples ( effective samples ) when the Markov chain reaches a stationary distribution. Now it is difficult to be triggered into a Bayesian network.
6. Reachability matrix refers to the degree that can be achieved after a certain length of paths between nodes of a directed graph is described in matrix form. The calculation method of reachable matrix is based on the operation properties of Boolean matrix.

 

The modified part of the file has been marked yellow. Looking forward to your reply.

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors evaluate the service performance of the loess railway tunnel. Based on the improved TOPSIS method, the indirect proximity degree of each risk factor was compared, and the appropriate service performance evaluation index was selected. Based on the ISM model and causality graph modification method, the dependency relationship between nodes was obtained, and the Bayesian network evaluation model was constructed. The EM algorithm was used for data learning by constructing the database, and the model was trained and verified. The overall accuracy (ACC ) and F1 value were used to evaluate the training and prediction effect of the model comprehensively. The paper is interesting and can be considered for publication in this journal. However, some shortcomings must be eliminated. The following list of comments should be helpful.

1. The methodological contribution should be emphasized (in the abstract and introduction - what is new in methodology?)

2. The acronyms must be first introduced, e.g.., ISM, and so on.

3. Each figure is another size. The figures have a very small font and are very hard to read.

4. The authors omitted the MCDA intro and review. Please present the background and current trends (methods) like COMET, SIMUS, SPOTIS, or DARIA-TOPSIS. It should be based on presenting why TOPSIS has been selected.

5. Some formulas are too big (font),

6. Presented approach should be compared with another one.

7. Future research directions should be presented in conclusion.

8. References should be refreshed with more literature from the last five years. 

 

Author Response

Thank you for your response and suggestions. Here is my reply one by one:

1. Please see manuscript for details. I have modified in the introduction part;
2. Please see manuscript for details;
3. Please see manuscript for details;
4. TOPSIS’s basic principle is to sort by detecting the distance between the evaluation object and the optimal solution and the worst solution. If the evaluation object is closest to the optimal solution and farthest from the worst solution, it is the best ; otherwise it is not optimal. Each index value of the optimal solution reaches the optimal value of each evaluation index. Each index value of the worst solution reaches the worst value of each evaluation index.
5. Please see manuscript for details;
6. This method is more applicable to the aspects studied in this paper. The results show that the established Bayesian network model has a high accuracy of 92%, which is easy to operate, effective and practical, and it is also applicable to the situation of incomplete index statistical data.
7. It has been modified in the conclusion part. Please see manuscript for details;
8. Please see the paper for details.

 

The modified part of the file has been marked yellow. Looking forward to your reply.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

I cannot agree with the answers to my key comment. The authors do not give a straightforward answer. They only say what TOPSIS is, not why it is better than SPOTIS, COMET, or DARIA-TOPSIS. Please add one or two paragraphs and answer them in the manuscript. Of course, the authors can take more methods to compare. However, this comparison and justification are crucial. Please improve it. The relevant literature on this topic should be easy to find by using keywords.

Author Response

Thank you for your response and suggestions.

It has been modified in Section 3.2 of the original text, including the advantages of TOPSIS, the comparison of TOPSIS with VIKOR method and CRITIC method. Amend as follows :

The advantages of this method are as follows : 1 simple method, reasonable struc-ture, clear order and flexible application ; 2 Make full use of the original data infor-mation, the ranking results can quantitatively reflect the pros and cons of different evaluation objects, intuitive and reliable ; ( 3 ) There is no strict requirement for data ( there is no strict limitation on data distribution type, sample content and index number, which is suitable for small sample data, multi-unit evaluation and multi-index large system, and is suitable for continuous and dynamic data ). It can be calculated directly with the original data without reducing the number of variables in the calculation process. 4 can eliminate the influence of different dimensions, so the evaluation indexes of different dimensions can be introduced at the same time for comprehensive evaluation.

Compared with the VIKOR ( VIseKriterijumska Optimizacija I Kompromisno Re-senje ) method, TOPSIS is a set of distances based on the distance from the ideal solution. The VIKOR method proposes a compromise scheme with a dominant rate based on the TOPSIS method. In the TOPSIS method, not only the nearest distance to the positive ideal solution, but also the longest distance to the negative ideal solution should be considered, so as to determine the optimal solution and maximize the efficiency. In addition, the process of TOPSIS method does not include any subjective factors, which is more suitable for the decision-making environment that requires completely objective results ; compared with the CRITIC ( Criteria Importance Though Intercrieria Correlation ) method, the CRITIC method loses some information due to the rank substitution of the index value, while the TOPSIS method can make full use of the original data information.

Based on this, this paper uses the improved TOPSIS method to screen and sort complex factors, simplify the model and reduce the amount of calculation.

The file is a modified version.

Round 3

Reviewer 2 Report

I appreciate the effort of the authors. However, I would like to point out that the COMET method has been juxtaposed with methods that do not quite fit in this direction. Also, additional explanations are required as to how the method is robust to ranking reversals. Why was this method not compared with SPOTIS and COMET and others previously suggested? Please address these issues in the manuscript. 

Author Response

Thanks for your reply. Please listen to my explanation : VIKOR and CRITIC, which were compared with TOPSIS in the last revision, are more methods used in the field of civil engineering. Secondly, we are familiar with them. The advantages of TOPSIS compared with them have been completed in the last revision. For example, ( 1 ) the method is simple, the structure is reasonable, the order is clear, and the application is flexible ; ( 2 ) Make full use of the original data information, sorting results can quantitatively reflect the pros and cons of different evaluation objects, intuitive and reliable ; ( 3 ) There is no strict requirement for data ( data distribution type, sample size, number of indicators are not strictly limited, suitable for small sample data, multi-unit evaluation, multi-index large system, suitable for continuous and dynamic data ). Without reducing the number of variables in the calculation process, the original data can be directly used for calculation. ( 4 ) The influence of different dimensions can be eliminated, so the evaluation indexes of different dimensions can be introduced at the same time for comprehensive evaluation.

For your given such as COMET, we have studied and understood, through the investigation found that it is widely used in the field of vocational education, in the field of civil engineering research and literature is less, so that we can not clearly compare their advantages. The same is true of several other methods. The less literature obtained from the survey is not enough to enable us to make an imprecise comparison. Therefore, we chose the methods we are familiar with and applied more in this field for comparison. Of course, we will continue to learn and understand the methods you mentioned, and hope to get your guidance and suggestions.

Author Response File: Author Response.pdf