Developing Two Hybrid Algorithms for Predicting the Elastic Modulus of Intact Rocks
Round 1
Reviewer 1 Report
In the present study, ANFIS-DE and ANFIS-FA models have been employed to predict the elastic modulus of intact rocks in which a data set is prepared from Azad and Bakhtiari dam sites, Iran. The research topic is interesting and the methodology is also acceptable, but the data description and results section needs an in-depth revision. The authors have to answer the comments carefully, apply and address in the main text.
1. The English of the paper is poor and deeds revision. It is recommended that the authors rewrite the entire paper.
2. The writing of the paper also needs significant improvement. There are many spelling and writing issues in the text of the manuscript that should be carefully reviewed. For example, Bakhtiari's name is written incorrectly (Bakhtiary is not correct). Also, there are typos in other parts (line 92 and ...).
3. A large amount of references have been provided without a detailed study and without a specific purpose (for example [6-18] and [19-33]). The author fail to properly cite the relevant references.
4. In order to general extend the literature review about the predicting models, following references are recommended.
- "Predicting resilient modulus of flexible pavement foundation using extreme gradient boosting based optimised models." International Journal of Pavement Engineering (2022): 1-20. DOI: https://doi.org/10.1080/10298436.2022.2095385
- "Machine learning-based intelligent prediction of elastic modulus of rocks at thar coalfield." Sustainability 14, no. 6 (2022): 3689. DOI: https://doi.org/10.3390/su14063689
- "Predicting the Young’s Modulus of Rock Material Based on Petrographic and Rock Index Tests Using Boosting and Bagging Intelligence Techniques." Applied Sciences 12, no. 20 (2022): 10258. DOI: https://doi.org/10.3390/app122010258
5. In lines 91 and 92, use the word "two" instead of 2.
6. The quality of the figures (especially figure 1) is very low and unacceptable.
7. In lines 82 to 84, use “section” instead of the “part”.
8. Sections 2.1 and 2.2 should be merged and presented in a more concise form.
9. Table 1 should be presented separately in the two parts of training and testing.
10.Dividing the data into two parts, 80% and 20%, respectively, for training and testing, was chosen on what basis? Different separations can lead to more accurate results.
11.In Tables 2 to 7, highlight or bold the best results.
12.The statistical graph of the data is poor. Histogram distribution charts (separately between training and test with different legends), box plot and Pearson correlation plots should be added to the manuscript.
13.The novel versatile and comprehensive performance evaluator indices such as PI, VAF and A10-index is strongly recommended to descript the model accuracy. They should merged in Table 8, in addition to used indices. Then, based on each index, the models should be scored and then ranked and discussed based on their accuracy. The best results in this table should be also bold or highlighted.
14.Convergence fitness curve against the R2 should be added and explained for each model in the paper.
15.The pairs of figures 8 to 11 should be individually merged so that the training and testing data should be displayed with different legends.
16.The sensitivity analysis using the best model (with the highest accuracy) considering the effect of removing each input variable should be presented and discussed. The related graph should be also drawn.
17.Taylor diagram should be presented for training and testing data set separately for each model.
18.For the developed models uncertainty assessment should be investigated.
Author Response
- Response to Reviewer #1
In the present study, ANFIS-DE and ANFIS-FA models have been employed to predict the elastic modulus of intact rocks in which a data set is prepared from Azad and Bakhtiari dam sites, Iran. The research topic is interesting and the methodology is also acceptable, but the data description and results section needs an in-depth revision. The authors have to answer the comments carefully, apply and address in the main text.
Dear Prof. / Dr.
We would like to thank you for reviewing our manuscript. Your comments are all valuable and helpful for revising and improving our paper. We have studied the comments carefully and have made corrections that we hope meet with your approval. We have studied the comments carefully and have made corrections that we hope meet with your approval.
- The English of the paper is poor and deeds revision. It is recommended that the authors rewrite the entire paper.
Response: Thank you for your comment. The English of the manuscript has been revised by a native speaker.
- The writing of the paper also needs significant improvement. There are many spelling and writing issues in the text of the manuscript that should be carefully reviewed. For example, Bakhtiari's name is written incorrectly (Bakhtiary is not correct). Also, there are typos in other parts (line 92 and ...).
Response: Thank you for your comment. The manuscript has been checked and revised.
- A large amount of references have been provided without a detailed study and without a specific purpose (for example [6-18] and [19-33]). The author fail to properly cite the relevant references.
Response: Thank you for your comment. The references have been updated.
- In order to general extend the literature review about the predicting models, following references are recommended.
- "Predicting resilient modulus of flexible pavement foundation using extreme gradient boosting based optimised models." International Journal of Pavement Engineering (2022): 1-20. DOI: https://doi.org/10.1080/10298436.2022.2095385
- "Machine learning-based intelligent prediction of elastic modulus of rocks at thar coalfield." Sustainability 14, no. 6 (2022): 3689. DOI: https://doi.org/10.3390/su14063689
- "Predicting the Young’s Modulus of Rock Material Based on Petrographic and Rock Index Tests Using Boosting and Bagging Intelligence Techniques." Applied Sciences 12, no. 20 (2022): 10258. DOI: https://doi.org/10.3390/app122010258
Response: Thank you for your comment. The above studies has been added to the manuscript.
- In lines 91 and 92, use the word "two" instead of 2.
Response: Thank you for your comment, and corrected.
- The quality of the figures (especially figure 1) is very low and unacceptable.
Response: Thank you for your comment. We have deleted figure 1, and added a new figure in this section.
- In lines 82 to 84, use “section” instead of the “part”.
Response: Thank you for your comment, and corrected.
- Sections 2.1 and 2.2 should be merged and presented in a more concise form.
Response: Thank you for your comment, and corrected.
- Table 1 should be presented separately in the two parts of training and testing.
Response: Thank you for your comment, and done.
- Dividing the data into two parts, 80% and 20%, respectively, for training and testing, was chosen on what basis? Different separations can lead to more accurate results.
Response: Thank you for your comment. Note that, the ratio of 80 and 20 for training and testing groups is suggested by many scholars such as Ye et al. [1], Fang et al. [2], Nguyen et al. [3] and Zhou et al. [4]. Aside from that, we have also tested the ratio of 70-30, nevertheless, the ratio of 80-20 had a better performance. Hence, we have used this ratio in our study.
[1] Ye, J.; Koopialipoor, M.; Zhou, J. et al. A Novel Combination of Tree-Based Modeling and Monte Carlo Simulation for Assessing Risk Levels of Flyrock Induced by Mine Blasting. Nat Resour Res 2021, 30, 225–243. https://doi.org/10.1007/s11053-020-09730-3
[2] Fang, Q.; Nguyen, H.; Bui, XN. et al. Estimation of Blast-Induced Air Overpressure in Quarry Mines Using Cubist-Based Genetic Algorithm. Nat Resour Res 2020, 29, 593–607. https://doi.org/10.1007/s11053-019-09575-5
[3] Nguyen, H.; Bui, XN.; Bui, HB. et al. Correction to: A comparative study of artificial neural networks in predicting blast-induced air-blast overpressure at Deo Nai open-pit coal mine, Vietnam. Neural Comput & Applic 2021, 33, 10615. https://doi.org/10.1007/s00521-021-05773-6
[4] Zhou, J.; Li, C.; Koopialipoor, M.; Jahed Armaghani, D.; Pham, B.T. Development of a new methodology for estimating the amount of PPV in surface mines based on prediction and probabilistic models (GEP-MC). International Journal of Mining, Reclamation and Environment 2021, 35:1, 48-68, https://doi.org/10.1080/17480930.2020.1734151
- In Tables 2 to 7, highlight or bold the best results.
Response: Thank you for your comment, and done.
- The statistical graph of the data is poor. Histogram distribution charts (separately between training and test with different legends), box plot and Pearson correlation plots should be added to the manuscript.
Response: Thank you for your comment. In this regard, three new figures have been added to the section 2.1.
- The novel versatile and comprehensive performance evaluator indices such as PI, VAF and A10-index is strongly recommended to descript the model accuracy. They should merged in Table 8, in addition to used indices. Then, based on each index, the models should be scored and then ranked and discussed based on their accuracy. The best results in this table should be also bold or highlighted.
Response: Thank you for your comment. The aforementioned performance evaluator indices, and also a total rank value, have been added to table 8.
- Convergence fitness curve against the R2 should be added and explained for each model in the paper.
Response: Thank you for your comment. Convergence fitness plan (R2) for each model have been presented in figures 13 and 14.
- The pairs of figures 8 to 11 should be individually merged so that the training and testing data should be displayed with different legends.
Response: Thank you for your comment, and done, as shown in figures 13 and 14.
- The sensitivity analysis using the best model (with the highest accuracy) considering the effect of removing each input variable should be presented and discussed. The related graph should be also drawn.
Response: Thank you for your comment. Based on your recommendation, the sensitivity analysis has been performed, and a new Table and figure have been added to the section 4.
- Taylor diagram should be presented for training and testing data set separately for each model.
Response: Thank you for your comment, and added (figure 16).
- For the developed models uncertainty assessment should be investigated.
Response: Thank you for your comment. Please consider our explanations.
In this paper, several artificial intelligence methods were developed to predict E, and their performances were then checked using several statistical indices. In the modeling, some effective parameters on the intensity of E, including depth of coring, density, porosity, durability, Poisson ratio, and P-wave velocity were used as the input parameters. According to obtained results, the ANFIS-FA models predicted the E values with the R2 of 0.991 and 0.988 in training and testing groups. These values clearly indicate that the performance of the proposed model was in a very high range, and the proposed model has a powerful reliability to predict E, and also has the capacity to generalize in other rock mechanics fields.
Best regards,
Corresponding author
Reviewer 2 Report
Dear Authors, Dear Editor
It was pleasure to review so well-organized and well-described paper. I do not have any serious remarks, just cosmetics which can improve the readability of whole paper:
- in line 50 it would be recommended to add citation (the the "variety of conditions" issue);
- it would be useful to add "Nomenclature" section. Number of symbols, acronyms is relatively high and it would make the understanding easier;
- improve the quality of schematic flowcharts (e.g. Figure 3, 4, 5) to improve readability;
- in tables 2-4, 5-7 just add column with relative discrepancy [%] of R2 to facilitate analysis of results;
- after Fig. 7 it would be better to add estimation error figure. Add also the error bars to measured data to show better agreement with experimental data;
One more time I would like to thank for review possibility of submitted paper.
Kind regards
Reviewer
Author Response
- Response to Reviewer #2
Dear Authors, Dear Editor
It was pleasure to review so well-organized and well-described paper. I do not have any serious remarks, just cosmetics which can improve the readability of whole paper:
Dear Prof. / Dr.
We would like to thank you for reviewing our manuscript. Your comments are all valuable and helpful for revising and improving our paper. We have studied the comments carefully and have made corrections that we hope meet with your approval. We have studied the comments carefully and have made corrections that we hope meet with your approval.
- In line 50 it would be recommended to add citation (the "variety of conditions" issue);
Response: Thank you for your comment. The proper references have been added.
- It would be useful to add "Nomenclature" section. Number of symbols, acronyms is relatively high and it would make the understanding easier;
Response: Thank you for your comment, and added.
- improve the quality of schematic flowcharts (e.g. Figure 3, 4, 5) to improve readability;
Response: Thank you for your comment, and improved.
- In tables 2-4, 5-7 just add column with relative discrepancy [%] of R2 to facilitate analysis of results;
Response: Thank you for your comment. Based on your recommendation to facilitate analysis of results, a new rank column has been added to these tables.
- After Fig. 7 it would be better to add estimation error figure. Add also the error bars to measured data to show better agreement with experimental data;
Response: Thank you for your comment. Based on your recommendation, a new figure has been added to the manuscript (figure 8).
One more time I would like to thank for review possibility of submitted paper.
Response: Thank you for your comment, and your positive feedback.
Best regards,
Corresponding author
Reviewer 3 Report
The manuscript focus on predicting the elastic modulus of intact rock. There are two elastic constants: Young modulus and Poisson's ratio. The presented method is probably good at predicting Poisson's ratio, too. This calculation should be interesting; however, even without it, the manuscript is okay.
Publishing the UCS values should also be useful.
Author Response
- Response to Reviewer #3
The manuscript focus on predicting the elastic modulus of intact rock. There are two elastic constants: Young modulus and Poisson's ratio. The presented method is probably good at predicting Poisson's ratio, too. This calculation should be interesting; however, even without it, the manuscript is okay.
Publishing the UCS values should also be useful.
Dear Prof. / Dr.
We would like to thank you for reviewing our manuscript and your positive feedback. For the future works, we will try to develop new models to predict UCS.
Round 2
Reviewer 1 Report
The authors have made considerable efforts. I appreciate them and I recommend this paper for publication.
