Intelligent Metal Welding Defect Detection Model on Improved FAST-PNN
Round 1
Reviewer 1 Report
Abstract
Real-time online detection of welding defects is already in place in the industry, while your method is time-consuming (to take photos, to analyse data, to predict the results etc.).
The last keyword has nothing to do with the article.
Introduction
Last paragraph, which should represent the originality of the article, has no verbs or actions, it is just an enumeration.
Also, in the article’s text, several paragraphs are simple listings, without verbs or actions.
2.3. It is unclear why you chose GLCM. There is no comparison with other methods, nor a plea for its advantages over them.
2.4. Why is FAST better? Make a comparison with normal PNN, discuss advantages or, in 3.5. present also the data analyzed using normal PNN.
Conclusions are just summarizing the article, without a conclusional idea.
Regarding the contributions: one author is only with the name, he did almost nothing for the article.
Lastly, I’m unconvinced that the article should be in the Coatings journal, because your work has nothing to do with coatings, not even with weldings themselves.
Author Response
Replies to the Reviewer
Thank you for your comments. We have revised and supplemented the manuscript carefully in accordance with the your comments.
We are very grateful for the insightful revision suggestions. For your comments, we have carefully addressed the issues, point by point. In accordance with the comments, we have made some corresponding changes in the revised manuscript. The changes are marked in Blue.
We have carefully checked the English grammar and language of the full text. The spelling and grammatical mistakes were revised.The spelling and grammatical review is marked in Red.
The details are as follows.
Comment(1)
Point 1. Abstract: Real-time online detection of welding defects is already in place in the industry, while your method is time-consuming (to take photos, to analyse data, to predict the results etc.)
Reply(1)
Thank you very much for your suggestion on the introduction section. We have carefully read the advice you provided and agreed that they were very helpful to our article. Please allow me to explain it to you: the early stage of building a welding defect detection model is time-consuming, but applying the trained model can achieve fast and accurate detection of welding defect damage types; At the same time, in order to further improve the detection efficiency of welding defects, we improved the traditional PNN network model and optimized the network structure. Based on the FAST-PNN model, the defect type can be quickly identified by simply inputting a picture of welding defects. At the same time, we've made changes in Blue, in Line 18-24 on Page 1, and in Line 385-388 on Page 14.
Comment (2)
Point 2. The last keyword has nothing to do with the article.
Reply (2)
Thank you very much for your suggestion.
We modified the last keyword, change the “ Improved Probabilistic Neural Network Model” to “FAST-PNN”, and marked in Blue in Line 26 on Page 1.
Comment (3)
Point 3. Last paragraph, which should represent the originality of the article, has no verbs or actions, it is just an enumeration.
Reply (3)
We agree with the comments.
We have revised the final conclusion and corrected some grammatical issues.
We have summarized the research methods of this paper, providing a certain theoretical basis and numerical reference for engineering practice. The relevant conclusions have been supplemented and revised in Blue in Line 374-388 on Page 14.
Comment (4)
Point 4. Also, in the article’s text, several paragraphs are simple listings, without verbs or actions.
Reply(4)
We agree with the comments.
We've made changes based on your suggestion in Line 180, Line 184, Line 190, Line 194 and Line 197 on Page 6-7.
The revision is marked in Blue.
Comment (5)
Point 5. 2.3. It is unclear why you chose GLCM. There is no comparison with other methods, nor a plea for its advantages over them.
Reply(5)
We agree with the comments.
Each welding defect is varied and complex. In order to better describe the changes of welding defects, the GLCM is selected to extract the texture feature parameters of welding defects. By adjusting the three construction factors (gray level g, generation step d and generation direction θ ) of the GLCM, 14 characteristic parameters can be obtained. It can describe the changes of welding defects in multiple directions and all angles. Therefore, the GLCM is selected as the method for quantitative analysis of welding defect images, and the network model of welding defect damage identification is constructed through the extracted characteristic parameters. We've made changes and explained your concerns in Line 96-103 on Page 3. The revision is marked in Blue.
Comment (6)
Point 6. 2.4. Why is FAST better? Make a comparison with normal PNN, discuss advantages or, in 3.5. present also the data analyzed using normal PNN.
Reply(6)
We carefully modified and added the FAST-PNN advantage in Line 171-174 on Page 5. And we added traditional PNN and FAST-PNN test data in Line 358-360 on Page 13-14. Thank you very much.
The revision is marked in Blue.
Comment (7)
Point 7. Conclusions are just summarizing the article, without a conclusional idea.
Reply(7)
We agree with the comments and make changes in the conclusion section. The revision is marked in Blue in Line 374-388 on Page 14, at the same time , the spelling and grammatical review is marked in Red.
Comment (8)
Point 8. Regarding the contributions: one author is only with the name, he did almost nothing for the article.
Reply(8)
We agree with the comments.
Regarding the author's contributions, we did not list them in detail; in fact, most of the projects were completed by two people. We redefine author contributions and mark Blue in Line 389-394 on Page 14.
Comment (9)
Point 8. Lastly, I’m unconvinced that the article should be in the Coatings journal, because your work has nothing to do with coatings, not even with weldings themselves.
Reply(9)
I think I will clear your concerns with my answer below:
In fact, one of the authors of this article specializes in material processing engineering and worked for a car seat manufacturing company (seat frame welding and forming plant) for three years. Another author studies image processing technology, intelligent recognition technology. The research of this paper is obtained by combining the work and research directions of the two authors. In fact, the research of this paper is to solve the problem of welding seam identification in the manufacturing site, which is related to the surface science and engineering of metal materials and the direction of metal materials in coatings magazine.
Thanks again for your advice.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
“The manuscript entitled ‘Intelligent Metal Welding Defect Detection Model on Im-2 proved FAST-PNN 3’ is within the scope of the journal Coatings. The paper is of sufficient scientific interest and has originality in its technical content to merit publication.
Issues to be addressed:
- In 3.1. (Line 209-210) it should be addressed which unfavorable factors in inspection environment effect on inspection.
- Figure 3 a and e are of poor quality. It is signed with crack for figure 3 a, but it seems more like porosity. Figure 3 e looks like end crater pipe which is not permitted imperfection.
- Figure 3 should have scale with a purpose to identify weld seam and defects size.
- Authors do not provide enough detail of the experimental methods so that the research cannot be replicable. It should be mentioned from how far, what angle and what magnification was used to acquire welding surface defect images. It should also be mentioned which material and welding process was used for investigation.
- Figure 5-1 and Figure 5-3: in legend should be all symbols used.
- In conclusion should be defined limits of sample space. How many images of individual defect is required to ensure accuracy and efficiency stated in results.
- Nowhere in article is mentioned about welding defect size identification method can detect. For example, weld on Figure 3 c is 100 % porosity. Based on that porosity sample, there is a question if identification method can recognize single pore on weld surface.
Author Response
Comments
The manuscript entitled ‘Intelligent Metal Welding Defect Detection Model on Improved FAST-PNN’ is within the scope of the journal Coatings. The paper is of sufficient scientific interest and has originality in its technical content to merit publication.
Replies to the Reviewer
Thank you for your comments. We have revised and supplemented the manuscript carefully in accordance with the your comments.
We are very grateful for the insightful revision suggestions. For your comments, we have carefully addressed the issues, point by point. In accordance with the comments, we have made some corresponding changes in the revised manuscript. The changes are marked in Blue. Spelling and grammar changes are highlighted in Red.
The details are as follows.
Comment(1)
In 3.1. (Line 209-210) it should be addressed which unfavorable factors in inspection environment effect on inspection.
Reply(1)
Thank you very much. We have added specific factors that affect the picture quality in the environment, such as light, welding smoke, oil on the surface of the weldment, etc., and eliminated the above factors affecting the picture quality through DIP technology. The revision is marked in Blue in Line 221-224 on Page 7.
Comment (2)
Figure 3a and e are of poor quality. It is signed with crack for figure 3 a, but it seems more like porosity. Figure 3e looks like end crater pipe which is not permitted imperfection; Figure 3 should have scale with a purpose to identify weld seam and defects size.
Reply (2)
Thank you very much for your suggestion. We have replaced Figure 3a and Figure 3e to allay your concerns; And all images are optimized and dimensioned. the revision is marked in Blue in Line 242-245 on Page 8.
Comment (3)
Authors do not provide enough detail of the experimental methods so that the research cannot be replicable. It should be mentioned from how far, what angle and what magnification was used to acquire welding surface defect images. It should also be mentioned which material and welding process was used for investigation.
Reply(3)
Thank you very much for your suggestion.
In Table 2 ,we have added and explained relevant welding parameters for the target weld to answer your concerns and marked in Blue in Line 225-228 , and in Line 234-235 on Page 7.
Comment (4)
Figure 5-1 and Figure 5-3: in legend should be all symbols used.
Reply(4)
Thank you very much for your suggestion.
Based on your suggestion, we have revised Figure 6-3 to ensure that the symbols used in Figure 6-1 and Figure 6-3 are consistent , in Line 352 on Page 13.
Comment (5)
In conclusion should be defined limits of sample space. How many images of individual defect is required to ensure accuracy and efficiency stated in results.
Reply(6)
Thank you very much. We increase the hint on the number of sample spaces.
The added content is marked in Blue in Line 283-285 on Page 9 and in Line 371-372 on Page 14.
Comment (7)
Nowhere in article is mentioned about welding defect size identification method can detect. For example, weld on Figure 3 c is 100 % porosity. Based on that porosity sample, there is a question if identification method can recognize single pore on weld surface.
Reply(7)
We agree with the comments.
We re-selected images with fewer pores in a single weld and placed them in Figure 3c in Line 244 Page 8; For relevant welding parameters such as welding dimensions, please refer to the content of Table 2 we added, in Line 228 on Page 7. The added content is marked in Blue.
Thanks again for your advice.
Author Response File: Author Response.pdf
Reviewer 3 Report
Dear Authors, Your Paper needs MAJOR REVISION and I strongly suggest to incorporate the below mentioned changes in Your Paper and to send the revised paper back to me:-
1. First of all, What is the welding methodology being adopted for joining the metals have not been mentioned in this paper. Introduction Part just simply states the metal weld processing methods. Authors must specify clearly “what type of welding methodology is being employed and from employment of which technology of welding, they got these weld defect samples”, in the Introduction section.
2. Introductory section does not give a clear picture about the need for carrying out this experimental analysis. I STRONGLY ADVICE the authors to read the below mentioned research papers thoroughly to get idea about how to strengthen Your “Introduction” Part:
· Dhanesh Babu, S.D., Sevvel, P, Senthil Kumar, R, Vijayan, V and Subramani, J: “Development of Thermo Mechanical Model for Prediction of Temperature Diffusion in Different FSW Tool Pin Geometries During Joining of AZ80A Mg Alloys”, Journal of Inorganic and Organometallic Polymers and Materials, 31 (7), 2021, 3196–3212. https://doi.org/10.1007/s10904-021-01931-4
· Stephan Thangaiah I S, Sevvel P, Satheesh C and Mahadevan S, “Experimental Study on the Role of Tool Geometry in determining the Strength & Soundness of Wrought AZ80a Mg Alloy Joints During FSW Process”, FME Transactions, Vol. 46, No.4, 612 – 622, 2018. https://doi.org/10.5937/fmet1804612T
· Dhanesh Babu SD, Sevvel P and Senthil Kumar R: “Simulation of heat transfer and analysis of impact of tool pin geometry and tool speed during friction stir welding ofAZ80A Mg alloy plates”, Journal of Mechanical Science and Technology, Vol. 34 (10), 2020, pp. 4239 – 4250. https://doi.org/10.1007/s12206-020-0916-7
Moreover, the references cited for justifying the 1st two paragraphs are very poor. The Authors are also STRONGLY ADVICED TO CITE THE ABOVE PAPERS in the Introduction Section.
3. In Chapter 2 entitled Method, the paper does not contain any photos of the welded samples, material being welded, parameters being adopted, machine (or equipment) employed for carrying out welding etc. Authors must compulsorily include the below mentioned details:
i. Description and properties of Material being welded
ii. Specifications (size, thickness etc.,) of the material plate or sheet being welded
iii. Photographs of welded samples
iv. Machine (or equipment) being employed to carry out the welding process
v. Welding Parameters being employed during the process
I STRONGLY ADVICE the authors to read the below mentioned research papers thoroughly to get idea about how to strengthen Your “Experimental Method” Part:
· Yokesh Kumar B and Sevvel P: “Impact of speed of traverse during joining of CDA101 plates by FSW process”, Scientia Iranica, 29 (4), 1817-1827 2022 https://doi.org/10.24200/sci.2022.58185.5612
· Yaknesh S, Sampathkumar K and Sevvel P: “Effect of Tool Pin Geometry and Process Parameters During FSW of Dissimilar Alloys of Mg”, Materials Research–Ibero–American Journal of Materials, Vol. 25. e20210508, 2022 https://doi.org/10.1590/1980-5373-MR-2021-0508
· Giridharan K, Sevvel P, Stalin B, Ravichandran M &Sureshkumar P: “Microstructural Analysis and Mechanical Behaviour of Copper CDA 101/AISI-SAE 1010 Dissimilar Metal Welds Processed by Friction Stir Welding”, Materials Research Ibero-american Journal of Materials, 25, 2022, e20210430. https://doi.org/10.1590/1980-5373-MR-2021-0430
4. In the caption of Fig.3 (c), correct the spelling of “Porostiyd”
5. Likewise, the paper contains many grammatical errors and authors are strongly advised to reedit the paper such that the grammatical mistakes are eliminated completely
Fig.3 (e) illustrates a normal defect free defect. But it is grouped as one of the figure under the caption “Grayscale images of five welding defects”. How Normal can be considered as a defect. Explain and justify clearlyComments for author File: Comments.pdf
Author Response
Comments
Dear Authors, Your Paper needs MAJOR REVISION and I strongly suggest to incorporate the below mentioned changes in Your Paper and to send the revised paper back to me.
Replies to the Reviewer
Thank you for your comments. We have revised and supplemented the manuscript carefully in accordance with the your comments.
We are very grateful for the insightful revision suggestions. For your comments, we have carefully addressed the issues, point by point. In accordance with the comments, we have made some corresponding changes in the revised manuscript. The changes are marked in Blue. Spelling and grammar changes are highlighted in Red.
The details are as follows.
Comment(1)
First of all, What is the welding methodology being adopted for joining the metals have not been mentioned in this paper. Introduction Part just simply states the metal weld processing methods. Authors must specify clearly “what type of welding methodology is being employed and from employment of which technology of welding, they got these weld defect samples”, in the Introduction section.
Reply(1)
Thank you very much for your suggestion. Based on your suggestion, we have added the metal welding process method and briefly described this method; The method is CO2 gas shielded arc welding. The revision is marked in Blue in Line 85-93 on Page 2.
Comment(2)
Introductory section does not give a clear picture about the need for carrying out this experimental analysis. I STRONGLY ADVICE the authors to read the below mentioned research papers thoroughly to get idea about how to strengthen Your “Introduction” Part:
- Dhanesh Babu, S.D., Sevvel, P, Senthil Kumar, R, Vijayan, V and Subramani, J: “Development of Thermo Mechanical Model for Prediction of Temperature Diffusion in Different FSW Tool Pin Geometries During Joining of AZ80A Mg Alloys”, Journal of Inorganic and Organometallic Polymers and Materials, 31 (7), 2021, 3196–3212. https://doi.org/10.1007/s10904-021-01931-4
- Stephan Thangaiah I S, Sevvel P, Satheesh C and Mahadevan S, “Experimental Study on the Role of Tool Geometry in determining the Strength & Soundness of Wrought AZ80a Mg Alloy Joints During FSW Process”, FME Transactions, Vol. 46, No.4, 612 – 622, 2018. https://doi.org/10.5937/fmet1804612T
- Dhanesh Babu SD, Sevvel P and Senthil Kumar R: “Simulation of heat transfer and analysis of impact of tool pin geometry and tool speed during friction stir welding ofAZ80A Mg alloy plates”, Journal of Mechanical Science and Technology, Vol. 34 (10), 2020, pp. 4239 – 4250. https://doi.org/10.1007/s12206-020-0916-7
Moreover, the references cited for justifying the 1st two paragraphs are very poor. The Authors are also STRONGLY ADVICED TO CITE THE ABOVE PAPERS in the Introduction Section..
Reply(2)
Thank you very much for your suggestion. We have carefully read and understood the paper you provided, at the same time, explained the mathematical modeling methods used in this paper. The revision is marked in Blue in Line 39-42 on Page 1, And the paper you provided is cited in the serial number [7-9].
Comment(3)
In Chapter 2 entitled Method, the paper does not contain any photos of the welded samples, material being welded, parameters being adopted, machine (or equipment) employed for carrying out welding etc. Authors must compulsorily include the below mentioned details:
- Description and properties of Material being welded
- Specifications (size, thickness etc.,) of the material plate or sheet being welded
iii. Photographs of welded samples
- Machine (or equipment) being employed to carry out the welding process
- Welding Parameters being employed during the process
I STRONGLY ADVICE the authors to read the below mentioned research papers thoroughly to get idea about how to strengthen Your “Experimental Method” Part:
- Yokesh Kumar B and Sevvel P: “Impact of speed of traverse during joining of CDA101 plates by FSW process”, Scientia Iranica, 29 (4), 1817-1827 2022 https://doi.org/10.24200/sci.2022.58185.5612
- Yaknesh S, Sampathkumar K and Sevvel P: “Effect of Tool Pin Geometry and Process Parameters During FSW of Dissimilar Alloys of Mg”, Materials Research–Ibero–American Journal of Materials, Vol. 25. e20210508, 2022 https://doi.org/10.1590/1980-5373-MR-2021-0508
- Giridharan K, Sevvel P, Stalin B, Ravichandran M &Sureshkumar P: “Microstructural Analysis and Mechanical Behaviour of Copper CDA 101/AISI-SAE 1010 Dissimilar Metal Welds Processed by Friction Stir Welding”, Materials Research Ibero-american Journal of Materials, 25, 2022, e20210430. https://doi.org/10.1590/1980-5373-MR-2021-0430.
Reply(3)
Thank you very much for your suggestion. We have carefully read and understood the paper you provided. About the 5 questions you mentioned about welding parameters, we have added Table2 (Target welding seam parameter table) and Figure 3(Welding parts drawing)to eliminate your concerns. The revision is marked in Blue in Line 225-233 on Page 7-8.
Comment(4)
In the caption of Fig.3 (c), correct the spelling of “Porostiyd”.
Reply(4)
Thank you for pointing out, we have changed “Porostiyd” to “Porostiy” in Red in Line 245 on Page 8.
Comment(5)
Likewise, the paper contains many grammatical errors and authors are strongly advised to reedit the paper such that the grammatical mistakes are eliminated completely
Fig.3 (e) illustrates a normal defect free defect. But it is grouped as one of the figure under the caption “Grayscale images of five welding defects”. How Normal can be considered as a defect. Explain and justify clearly.
Reply(5)
Thank you very much for your suggestion. We have carefully checked and corrected the full text for grammar and spelling errors. Spelling and grammar changes are highlighted in Red.
In order to better distinguish between normal and defective welds, we also incorporate normal welds into the target welds for weld detection and identification, which is more conducive to engineering practice applications, The revision is marked in Blue in Line 238-241 on Page 8.
Thanks again for your advice.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Now the text is improved and can be published with some minor English improvements still necessary. The Editor will contact you about that.
Reviewer 3 Report
Dear Authors,
I appreciate the changes and revisions being made in the revised Manuscript. I recommend the revised paper for publication in this Journal.
