Study on Accuracy Metrics for Evaluating the Predictions of Damage Locations in Deep Piles Using Artificial Neural Networks with Acoustic Emission Data
Round 1
Reviewer 1 Report
Thank you for giving me the opportunity for review. Authors presented interesting work on accuracy metrics evolution of ANN model on damage locations in deep piles. I recommend some minor revisions before its acceptance
- An introduction section, it is important to present a literature review of the previous works
- Are there any specific reasons to choose mentioned seven performance parameters of ANN models?
- Could you elaborate or discuss more 6 trained models? Abbreviations for them were missing
- How this training has been conducted to develop a model and how about validation
- Please conduct English revision with a native speaker
Author Response
The authors are very grateful for the honorable reviewer’s comments.
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
Reviewer report Applied Sciences
Paper: Studies on Accuracy Metrics for Evaluating the Predictions of Damage Locations in Deep Piles Using Artificial Neural Network with Acoustic Emission Data
General Comment
The topic of the current paper is interesting as ML algorithms are becoming more and more popular.
The paper is well written with a simple language, however as general comment I ask to the authors to make a careful revision of the repetition in the text and to correct the grammar and spelling errors.
The second general comment is that the quality of all the plots is very la very low so that the number and label in the figures are not readable. I ask to enlarge and improve the quality of all the figures of the manuscript.
Then as general comment it is not clear in what consists the method of multi-installation described in group 1 as the label of group 1 is not visible in Figure 2 and the combination to build this group is not explicitly explained in the text. I suggest the authors to extend its explanation in the section materials and methods.
Minor comments:
Pages 4-6 – To much time is repeated the sentence:”variables refer to the distance from ground level in this case study.”. I suggest to report it one time at the beginning of the analysis.
Page 7- lines 2. correct 0.5 m
Page 7 – lines 3:” there, have” very likely a different word than there could be more comprehensible here.
Page 8 – after figure 4: correct: illustrated and correct: according. Check the spelling of the whole page.
Table 1 - I would suggest to use bold and italic to show the max and min values in Table 1 as described in the text
Page 9: correct: TRAINGLM in the text as referred in Table 1.
Figure 5 is too small and at low resolution and it is very difficult to read both x axis and y axis values and labels on the histograms. In addition, the red line is not very visible as its labels.
Section 4.1.2- first line: correct: performance and correct: percentage-dependent; correct: illustrated. Check the spelling of the whole page
Page 10: please correct the name of the algorithm TRAINGLM as in table or in the text
Page 10: correct: “zeros were removed”. Check the spelling of the whole page
Page 11: still not clear if the name of the algorithm is TRAINLM as in the main text or as that reported in Table 1 and 2.
Page 11: it is not clear why the total dataset in group 1 is the reference for the subset of data that is group 2 and group 3. Please explain better in the text. Please explain which type of combination you have in group 1.
Figure 6, 7 and 8 need to be redrawn with higher quality and in larger dimension to be understandable and visible (same for figure 5).
Page 11- last line before the table 3 - correct: group 3> group 1> group2
Section 4.2.2 -line 1: correct the verb in: illustrate
Table 4 – description of groups - please correct group 3 instead of the second label group 2.
Figure 9 and 10 are also of very poor quality and need to be improved
Page 14 – it is needed to introduce a short explanation of the meaning of the output of the CV index before to discuss the results, i.e. higher CV values highlight results more sensitive to the predictions.
Finally also figure 11 is of very low quality and needs to be redraw.
Section 6. conclusion - please deeply revise the conclusion to avoid multiple repetition of the same words and verbs. This is a general comment along the whole manuscript, please select more carefully the terms in order to improve the language level. In the conclusion please do not quote figure 11 but rather explain it in the sentence.
Comments for author File: Comments.docx
Author Response
The authors are very grateful for the honorable reviewer’s comments.
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 3 Report
-please add colorful picture of measurements (optionally)
-please add block diagram of the proposed research step by step ;;; what is the result of paper?;;;
-please add block diagram of the proposed method;;;
-please add photo/photos of application of the proposed research ;;;;
-figures - please add labels, describe it better.
-please add sentences about future analysis;;;
-references should be 2018-2021 Web of Science about 50% or more ;;
It seems that it is ok but please check it.
-Please compare with other methods, justify. Advantages or Disadvantages;;;
for example:
1) Acoustic fault analysis of three commutator motors, Mechanical Systems and Signal Processing, vol. 133 art. no. 106226, 2019,
https://doi.org/10.1016/j.ymssp.2019.07.007
-Conclusion: point out what are you done;;;;
-Could you use the proposed method for other acoustic signals?
Author Response
The authors are very grateful for the honorable reviewer’s comments.
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 4 Report
In this paper the authors study the Accuracy Metrics for Evaluating the Predictions of Damage Locations in Deep Piles Using Artificial Neural Network with Acoustic Emission Data
Section 1 can be improved. Before introducing the ANNs, it would be appropriate to better describe the underlying problem and clarify how the acoustic signals are used to monitor the health of the foundation piles. At the end of the section, add an outline of the rest of the paper, in this way the reader will be introduced to the content of the following sections.
Section 2 can be improved. You have to clarify which correlation coefficient you are referring to, also better understand the terms of the equations. Correct the format of the equations, I have inserted specific advice for each page.
Section 3 must be improved. The most important thing is that the technical specifications of the AE sensors are totally missing. What is it about? I suppose these are accelerometers, in this case you could talk about vibrations, otherwise you should refer to microphones. This is a serious lack, these are the tools with which you collect the data that you will then use for the evaluation of the metrics so you must be clear and precise. Indicates the type of sensor, accuracy, sensitivity, etc. Indicate how they were connected to the concrete structure, also indicate how you recorded the data and with which software you processed them. It would be appropriate to organize all this information in a table so that the elector can more easily retrieve such data. You also make no mention of the type of damage you are trying to identify. Add photos of the damage and sensor installation so the reader can understand what we are dealing with. You need to spend more time describing the ANN as it is an essential part of your job.
Section 4 must be improved. If you decide to use 6 different training algorithms it is necessary that you describe them in detail and add appropriate references, otherwise readers will not be able to follow the flow of information. Improve the quality of the figures according to the suggestions that I have specified in the following sections.
Section 6 must be improved. Describe your assumptions more carefully, then summarize the results obtained, finally indicate which possible uses of the methodology can be implemented. Paragraphs are missing where the possible practical applications of the results of this study are reported. What these results can serve the people, it is necessary to insert possible uses of this study that justify their publication. They also lack the possible future goals of this work. Do the authors plan to continue their research on this topic?
Unfortunately the pdf that was provided to me did not have line numbers so the precise references were more complicated.
Page 2) “A value of +1 (or -1) indicates the perfect correlation between two variables” Explain the difference between 1 and -1 otherwise it seems that the same result is obtained. Do the same for the other values as well.
Page 3) What type of correlation coefficient are you referring to? Specify it and also indicate what other types there are (Pearson, Spearman). Specify that equation (1) refer to Pearson’s correlation coefficient. In equation (2) specify that the numerator is the sum of squares of residuals also called the residual sum of squares and the denominator is the total sum of squares that is proportional to the variance of the data. Enhance the caption of Figure 1, add some descriptive content of the images you added in order to explain the correlation between the variables.
Page 4) Improve the format of equations (3), (4) and (5), these are too too close, leave a line between the equations, also try to show the summation in extended form with the index above and below.
Page 5) Improve the format of equations (7) and (8), these are too too close, leave a line between the equations, also try to show the summation in extended form with the index above and below. Finally he leaves a blank line before and after the equations.
Page 6) It is not clear how the sensors are divided into the three groups, in reality I only see two groups (group 2 and group 3). I don't understand what group 1 is about. The technical specifications of the AE sensors are totally missing. What is it about? Figure 2 needs much improvement. First of all it is better to refer to the pole section and not 2D sketch. Also add a legend in which you indicate the descriptors of the sensors and damage, also improve the quality of the image use at least 300 dpi. Finally improve the caption add information, think that the reader should find all the information to be able to read the image here. You also make no mention of the type of damage you are trying to identify. Add photos of the damage and sensor installation so the reader can understand what we are dealing with.
Page 7) You say it's 5 points but in Figure 2b I see 15. Try to be clear in the presentation of your set up. You need to better describe how the 100 signals were generated, which tool did you use? Talk about hits with what were generated. This part is weak, you must necessarily strengthen it. An ANN backpropagation can also have more than one hidden layer, this is not what characterizes this system architecture.
Page 8) Introduce adequately the six train algorithms.
Page 9) Figure 5 must be improved. You have put many things that are repetitive, you can make a single figure in which you insert all the curves and bars. Remove the values of R because otherwise you have to add the values of R2 as well. Use different descriptors for the three curves of the scale-depended metrics
Page 10) Figure 6 must be improved. You have put many things that are repetitive, you can make a single figure in which you insert all the curves and bars. Remove the values of R because otherwise you have to add the values of R2 as well. Use different descriptors for the two curves of the percentage-dependent metrics.
Page 12) Figure 7 must be improved. You have put many things that are repetitive, you can make a single figure in which you insert all the curves and bars. Remove the values of R because otherwise you have to add the values of R2 as well. Use different descriptors for the three curves of the scale-depended metrics
Page 13) Figure 8 must be improved. You have put many things that are repetitive, you can make a single figure in which you insert all the curves and bars. Remove the values of R because otherwise you have to add the values of R2 as well. Use different descriptors for the two curves of the using percentage-dependent metrics
Page 14) Improve the format of equation (11)
Author Response
The authors are very grateful for the honorable reviewer’s comments.
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 2 Report
Dear Authors,
Thanks for the work done to reply to my comments and to improve the revised version of your work.
I am satisfied with this revised version. Therefore I suggest the acceptation of this paper as it is.
Author Response
The authors are very grateful for the honorable reviewer’s comments. The respected reviewer’s valuable comments have played very important role for improving our paper. Many thanks to the respected reviewer again.
Reviewer 3 Report
-it is good idea to add photo of application for example fault diagnosis of machine or something
-graphs it is good idea to add labels what metrics are on axes OX and OY, reader should know in 3 seconds what is what in the figures.
-formulas should be formatted
Author Response
The authors are very grateful for the honorable reviewer’s comments.
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 4 Report
I state that the version received by the reviewer lacked line numbers, so it was more difficult to check precisely the changes made and specify which part of the paper needed to be improved.
The authors addressed all the reviewer's comments with sufficient attention and modified the paper consistently with the suggestions provided. The new version of the paper has improved significantly both in the presentation that is now much more accessible even by a reader not expert in the sector, and in the contents that now appear much more incisive.
Minor revision:
Page 2) You could improve the references to references, enter the number immediately after mentioning the authors. For example: William et al. [3]
Page 7) Enhance the information of the AE sensor this is the sensor with which you acquire data. Mention the manufacturer (Physical Acoustics Corporation) and then add Peak Sensitivity, Operating Frequency Range, Resonant Frequency, Directionality. As already advised, you could put them in a table. You could add that a more representative photo, in figure 3h you can't quite appreciate it.
Page 8) Do not use abbreviation such as i.e. Check throughout the paper
Page 10) Table 1. It is necessary to add references to the algorithms listed in the table to allow the reader to deepen the topic.
Author Response
The authors are very grateful for the honorable reviewer’s comments.
Please see the attachment.
Author Response File: Author Response.docx
