Depth and Angle Evaluation of Oblique Surface Cracks Using a Support Vector Machine Based on Seven Parameters

Round 1

Reviewer 1 Report

This manuscript proposes a novel method to analyze results from nondestructive detection of cracks using laser ultrasound technique, which is suitable as a topic for the journal, applied sciences. A support vector machine used in the present method is one of neural network models widely used in AI field which attracts a lot of attention from readers in broad research areas. In the present method, SVM was successfully used to evaluate geometrical information of cracks embedded in a material such as depth and angle from limited results of the laser ultrasound system. 

To effectively train a neural network model, parameters related to desirable results have to be appropriately chosen and a lot of data of that kind are required to be provided with the model that is trained. In my opinion, one of the most important point in this article is that a variety of data was made available to select appropriate feature parameters by conducting finite element analyses which model thermal and acoustic response to the laser ultrasound source. Verification results on real experiments successfully demonstrated that the effectiveness of the optimized seven feature parameters as mentioned in the title.

In my opinion, from these significance of the results, this article should be published basically as it is. However, I propose some minor points to be fixed for making readers better understand as follows.

Line 209
In table I, lame parameter μ is missing.

Paragraph from line 210
I think it is better to add a figure to describe geometry of a FEM model such as location, depth and angle of a crack as well as boundary conditions. One of 150 samples is sufficient to be described in the figure. This will make readers better understand not only this paragraph but also description of results from line 237.

Line 254
I think it is better to use same time range for Figure 4(a) and (b) to understand corresponding location of peaks in (a) and (b) in a whole analysis.

Line 380
Same font is better to be used for subtitles of Figure 12 (a) and (b).

Line 472
"Correlation efficient" may be "Correlation coefficient"

Line 582
Figure 21 may be Figure 22.

Line 649-
I think it will look better if same indent is applied to all lines for each point of conclusions.

Line 663-
I think it will look better without space between lines.

 

Thank you for your attention.
Best regards,

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper presents a SVM method to identify Oblique Surface Cracks. The paper is well organized and written. Some minor problems remains to be published.

 

1) The paper is too long. The page of the paper is 25. Basic principles of SVM can be removed in the paper.

 

2) Please check equations such as (8) and (9). dot (T) is not proper. Also, In Table I, second Lame constant "mu" is missing

 

3)  It seems that Figure 3 means the displacement, but Legend in Figure 3 is missing. For other figures, legends are also missing.

 

4) The caption of the Figure 16 is confusing. Please re-write the captions for clear understanding. 

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf