Ensemble Deep Learning Ultimate Tensile Strength Classification Model for Weld Seam of Asymmetric Friction Stir Welding
, Rapeepan Pitakaso 1, Kanchana Sethanan 2, Monika Kosacka-Olejnik 3, Thanatkij Srichok 1,* and Peerawat Chokanat 4
Round 1
Reviewer 1 Report
The paper is well written and organised. The research results can be of great interest to the relevant researchers. Thus, the paper can be accepted by the journal.
Author Response
We greatly appreciate your insightful comment; we will continue to conduct excellent future study.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The work of this paper is of interest, but the manuscript needs some work for publication.
1. Please note that the Traditional UTS of joints is usually determined by tensile testing.
2. Please explain how to identify the surface characteristics of the FSW joint.
3. Please explain how FSW process parameters are determined based on deep learning.
Author Response
Thank you very much for your insightful comment; we updated our manuscript and provide the full point-by-point response in the attachment. Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
In general, this is an interesting topic, which uses the image processing for predicting the quality of friction stir welding. However, authors have to check these issues:
- Figure 1: The article name mention about the tensile strength. However, figure 1 shows the bending test. I think this is not proper in meaning. So, authors have to check this issue
- Table 1: please explain about the selection parameters like: reason for selecting, effect of them on the weld quality / weld geometry,...
- Line 210-211: please describe about the experiment model. Example: dimension of sample, position for taking the welding picture, steps for processing the experiment data,…
- Table 2 should have the unit of strength
- Figure 5 has to be mentioned in paragraph
- Table 5: please define the CNN1,2,3,4,5
- The accuracy (Table 6) should be defined
- Table 9: 10 last model use the picture with or without segmentation?
- Table 9: Do you have any explanation about the higher accuracy of segmentation case (0.9189)?
- In my opinion, part 5 (discussion) should be imported in to part 4 (result). So, the discussion or explanation will appear after each result. However, this structure is still accepted.
- In addition, this version mention clearly about the deep learning steps, methods,… However, the experiment and the testing steps are also should be described clearly, as well as the comparison between the prediction and experiment result should be show out clearly.
Author Response
Thank you very much for your insightful comment; we updated our manuscript and provide the full point-by-point response in the attachment. Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 4 Report
This work is a timely effort by the authors on "Ensemble Deep Learning Ultimate Tensile Strength-Classification Model for Weld Seam of Asymmetric Friction Stir Welding". However, there are few sugessions to improve the quality of presentation of this manuscript.
1. Novelty needs to be highly improved.
2. The authors must add more explanation in the material and methods on Friction Stir Welding (FSW) process and how they took the 1664 images of the final strength of joint.
3. There are few references which are old too which must be repalced with those appeared in last five years.
4. They must mention the software tools used to analyse the deep learning method for any process with all possible neural networks.
5. The discussion section must be elaborated to a greater extent. Currently, the density of all other written stuff within manuscript is sufficiently higher than that of discussion section.
6. Tables showing accuracy must be discussed in a better way and how the values in those tables were calculated.
7. There must be some explained work on the validation of the methodology at the end with certain FSW factors without having tested it for UTS.
Author Response
Thank you very much for your insightful comment; we updated our manuscript and provide the full point-by-point response in the attachment. Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 3 Report
This version was bright and easy for understanding the research content. Especially the result in Table 12.
Therefore, I suggest that this version could be published.
Sincerely yours,
Reviewer 4 Report
None
