Possibilities of Artificial Intelligence-Enabled Feedback Control System in Robotized Gas Metal Arc Welding
Round 1
Reviewer 1 Report
Reviewer comments
In this paper, A back propagation neural network algorithm is used for optimizing GMAW process parameters. It is found that the intelligent welding system can be utilized in welding feedback control with multiple welding conditions and welding process parameters, which has a great advantage for developing the intelligent welding system. At the same time, this paper is clearly written and has complete experiment results. So this manuscript can be accepted and published under minor modifications. The modifications are shown below:
1、Why choose Mean Squared Error (MSE) as the network performance evaluation, Please the author makes some detailed explanations.
2、In this paper, the adjustment of welding parameters is realized by BP neural network. How to consider the delay between data transfer and calculation speed in the actual welding process, Please the author makes some detailed explanations.
3、The actual welding torch TCP in this paper has a large distance from the position of the sensor, and how to realize real-time feedback of data.
4、The model error in Fig 6 is large, Has the author optimized the model? It is suggested that the author give the specific value of the error between the test set and the training set.
5、What is the data collection of the sample? It is recommended that the author use the schematic diagram to describe it. At the same time, It is recommended to give a detailed description and schematic diagram for evaluating the weld quality.
Author Response
Please see the attachment
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This paper discusses the possibility of applying artificial intelligence feedback systems in automated gas shielded welding. The presentation of the experimental part of the paper is weak and does not provide various forms of data presentation. Analyzing the algorithm alone is obviously not enough. On the basis of the lack of innovation in the algorithm, it is necessary to clarify what improvements the algorithm has brought to the processing results. Several suggestions have been provided, and the author can refer to them for further optimization in future research work.
1. The abstract of the paper should briefly and clearly display the work and significance of the author. It is suggested to reduce the description of the background and highlight the innovative points of the author's work and what it has brought to the upgrading of this processing technology.
2. It is suggested to present the data format generated by the experiment and provide the dataset used for training.
3. It is suggested to display the predicted results instead of only evaluating the algorithm.
4. It is suggested to conduct more detailed experimental verification on the improvement of final welding quality.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Reviewer 3 Report
- Add a brief comment in the "Introduction" paragraph concerning the role of the Numerical Modeling on the welding process, in order to give a more detailed description of the current state:
1. https://doi.org/10.1007/s00170-021-08401-8
2. https://doi.org/10.1016/j.ijheatmasstransfer.2023.124114
3. https://doi.org/10.1016/j.mtcomm.2022.105280
- It is not clear the welding parameters used for the experiments. Add a table in order to make it clear for the reader;
- It was not clearly explained the investigated parameters establishing the good quality of the welding;
- It was not considered the repeatability of the experiments. It seems the experiments were performed only one time, and it is not corrected from an experimental point of view.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Reviewer 4 Report
This article tried to use ANN to control the welding quality of the welded joints produced by gas metal arc welding. This work seems appealing. However, there are a couple of issues needed to be made clear of.
1). Line 89. I guess it will be better if the details of the materials, welding wire, shield gas and so on are added to this part.
2). Figure 3. The groove profile was measured. Would you mind showing an example of the measured groove profile and the real profile to present more details.
4). Line 154. Please list a table to show the details of the Taguchi design (the welding process parameters, the range of the parameters, the results and so on).
5). Line 190, what are the inputs and outputs of the NN? Please specify the inputs and outputs of NN.
6). Line 210. Please specify the amount of data for training, testing and validation.
7). Line 210. Please list some statists features of the data you used.
8). Line 268. “Thus, the most suitable network configuration for this case is the 2-20-20-2 network.” Please specify the inputs and outputs of the NN.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Reviewer 5 Report
Intelligent welding system for feedback controlling of the gas metal arc welding was developed.
The parameters of the root height and root gap were used for feedback controll.
The intelligent welding system can be used for his “own” welding equipment.
Adequate methods were used, the reached results are adequate presented.
The paper is well written, the text is clear and easy to read.
The conclusions are in line with the evidence and arguments presented.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
The paper has undergone significant revisions, making the content more rich and relevant. Several comments have been made on the revised paper for the author's reference.
1. It is recommended to display Figure 2 in conjunction with the physical image to make the display results more comprehensive and clear.
2. It is recommended to modify some modules into images in the block diagrams of Figures 3 to 5 for easier understanding and reading.
3. It is recommended to interpret and analyze the data in Figure 6, as well as which part of the content has been verified.
Author Response
Dear reviewer, thank you for your comments and corrections suggestions for improving the paper. We found the comments well fit and by improving the aspects, the quality of the paper is increased. We have made the corrections based on your comments, we hope that you find the corrections and clarifications satisfactory. We added the answers to your comments under the comment in form of bullet points.
- It is recommended to display Figure 2 in conjunction with the physical image to make the display results more comprehensive and clear.
- Unfortunately, we did not take physical pictures of the groove before the welding. The figure of measurement before the welding is demonstrated in the figure 1. The same data from the figure 1 scan is visualized in the figure 2. This was clarified in more detail in the text to better explain the measurement process.
- It is recommended to modify some modules into images in the block diagrams of Figures 3 to 5 for easier understanding and reading.
- Figures 3 and 4 were modified accordingly. However, figure 5 was left as it was as the visual images tended to expand the figure extensively and reduce the readability.
- It is recommended to interpret and analyze the data in Figure 6, as well as which part of the content has been verified.
- The content in figure 6 was analyzed and the content of it was discussed in more detail. The figure 6 content was completely verified and used as a training data. This was also clarified in the text.
Reviewer 4 Report
This manuscript has improved a lot after revising. However, there still exist some small errors needed to be revised.
1). Line 108, Table 1. It seems that some items repeat in this table.
2). Please provide the chemical compositions and mechanical properties of the welded steel and the filler material.
Author Response
Dear reviewer, thank you for your comments and corrections suggestions for improving the paper. We found the comments well fit and by improving the aspects, the quality of the paper is increased. We have made the corrections based on your comments, we hope that you find the corrections and clarifications satisfactory. We added the answers to your comments under the comment in form of bullet points.
1). Line 108, Table 1. It seems that some items repeat in this table.
- The repeating items in the table were corrected.
2). Please provide the chemical compositions and mechanical properties of the welded steel and the filler material.
- The chemical composition and mechanical properties of the base material and filler wire were added in table 2.
