Intelligent Monitoring and Compensation between EDM and ECM

Round 1

Reviewer 1 Report

1. The sections like Abstract, Introduction and Conclusions should be simplifed. Some common introduction about ECM ans EDM should be deleted.

2. Please check the manuscipt carefully because of some grammar mistakes. For example, line 52 to line 56 on page 2.

3. The x-coordinate like Fig.4，Fig.15 and Fig.16  should be redrawn.

4. In normal ECM, the voltage is set constant. So fig.5 could be deleted.

5. Scale bar should be labelled in fig.13.

6. Fig.15 and Fig16 should be plotted in sub-figures with the same x coordinate.

Author Response

Reviewer 1:

1. The sections like Abstract, Introduction and Conclusions should be simplifed. Some common introduction about ECM ans EDM should be deleted. Please check the manuscipt carefully because of some grammar mistakes. For example, line 52 to line 56 on page 2.

Thank you very much for suggestion. We have revised all the manuscript in accordance with the reviewer’s comment as follow:

Electrochemical machining(ECM) is a machining method. It has the advantages of highly cost-effective, high-strength, and it's easy to process the heat-resistant materials into complex shapes, but this process is vulnerable to the influence of material and workpiece structure, the hydrodynamic instability of the anode boundary layer affects the surface roughness

3. The x-coordinate like Fig.4，Fig.15 and Fig.16  should be redrawn.

Thank you very much for suggestion. We have revised all the manuscript in accordance with the reviewer’s comment as follow:

 

Figure 4. Current value of five experiments of ECM

 

Figure 15. Average spark frequency

 

Figure 16. Average ignition delay time

 

4. In normal ECM, the voltage is set constant. So fig.5 could be deleted.

Thank you very much for suggestion. In Figure 6(a) in the article, it can be seen that there is a grouping effect, and increasing the voltage can effectively improve the identification of the model.

5. Scale bar should be labelled in fig.13.

Thank you very much for suggestion. We have revised all the manuscript in accordance with the reviewer’s comment as follow:

 

Figure 13. Comparison before and after processing

 

6. Fig.15 and Fig16 should be plotted in sub-figures with the same x coordinate.

Thank you very much for suggestion. We have revised all the manuscript in accordance with the reviewer’s comment as follow:

 

 

Figure 15. Average spark frequency

 

Figure 16. Average ignition delay time

 

 

Author Response File: Author Response.docx

Reviewer 2 Report

Dear Authors,

the article is interesting and has publication potential. In order to improve it, I suggest making some corrections:

In the abstract, it looks like there are too many spaces between sentences.

Lines 35-39 - the sentence is too long and therefore stylistically incorrect.

Lines 40-41 - After nickel-based some word is missing. Nickel-based what?

After reading the first page, I conclude that the article needs intensive stylistic corrections, as well as the English language.

Lines 41-41 - what causes frustration in aerospace?

Line 63 - What technical difficulties are associated with this process?

Lines 75-97 - In the description of the EDM process, please include the key feature describing the process - discharge energy. It has an important influence on shaping the roughness (topography) of the surface. Examples of publications describing discharge energy can be found, for example, in:

https://doi.org/10.1016/j.jmatprotec.2021.117083

https://doi.org/10.3390/cryst11111371

https://doi.org/10.3390/coatings9110718

https://doi.org/10.1016/j.triboint.2021.107139

https://doi.org/10.1016/j.jmatprotec.2003.10.036

Figure 1 - Are the current and voltage in the EDM sufficient to infer the accuracy of the part? What about the duration of the pulse, the gap between the electrode and the workpiece? Are the current and voltage in the EDM sufficient to infer the accuracy of the part? What about the duration of the pulse, the gap between the electrode and the workpiece?

The description of the neural network used is not sufficient. Please describe the activation functions used, the network learning algorithm used.

Line 201 - Why, despite having 200 examples of results, only 20 were used to train the network? The fact that this number was automatically cut is not a sufficient explanation.

Improve drawing resolution. Especially picture number 9.

The article is generally good, but needs a lot of refinement to be published. Most of these corrections are simply careless and hasty drafting of the paper. Please be more accurate with the text, figures and graph descriptions for this publication.

Author Response

Reviewer 2:

1. In the abstract, it looks like there are too many spaces between sentences.

Lines 35-39 - the sentence is too long and therefore stylistically incorrect.

Lines 40-41 - After nickel-based some word is missing. Nickel-based what?

After reading the first page, I conclude that the article needs intensive stylistic corrections, as well as the English language.

Lines 41-41 - what causes frustration in aerospace?

Thank you very much for suggestion. We have revised all the manuscript in accordance with the reviewer’s comment as follow:

Development new materials and process are still many problems and frustrations in the aerospace, which makes it difficult to improve the effi-ciency, so this technology is worth investing in development.

2. Line 63 - What technical difficulties are associated with this process?

Thank you very much for suggestion. We have revised all the manuscript in accordance with the reviewer’s comment as follow:

“Accuracy and surface roughness are still technical difficulties that have not been overcome.”

 

3. Lines 75-97 - In the description of the EDM process, please include the key feature describing the process - discharge energy. It has an important influence on shaping the roughness (topography) of the surface. Examples of publications describing discharge energy can be found, for example, in:

https://doi.org/10.1016/j.jmatprotec.2021.117083

https://doi.org/10.3390/cryst11111371

https://doi.org/10.3390/coatings9110718

https://doi.org/10.1016/j.triboint.2021.107139

https://doi.org/10.1016/j.jmatprotec.2003.10.036

Thank you very much for suggestion. We have revised all the manuscript in accordance with the reviewer’s comment as follow:

“In the EDM process, discharge energy has an important influence on shaping the roughness and topography of the surface. Investigate the influence of the powder particles and capacitor connected parallel in EDM and especially in the gap, on the thermal spread in the dielectric and on the influenced zone. A bigger workpiece surface has greater discharge energy distributed [21,22]. The features directly related to the shape in the texturing process in relation to the discharge energy and the contact angle [23].

”

4. Figure 1 - Are the current and voltage in the EDM sufficient to infer the accuracy of the part? What about the duration of the pulse, the gap between the electrode and the workpiece? Are the current and voltage in the EDM sufficient to infer the accuracy of the part? What about the duration of the pulse, the gap between the electrode and the workpiece?

The voltage and current of EDM can be estimated after Feature calculation. For example, the ASF can know the processing environment of electrodes and workpieces when the spark is presented.

5. The description of the neural network used is not sufficient. Please describe the activation functions used, the network learning algorithm used.

Thank you very much for suggestion. We have revised all the manuscript in accordance with the reviewer’s comment as follow:

“The feature calculated by EDM is the input of the NN model, so as to establish a NN model, as shown in Figure 17(a), to estimate the processing quality, The analysis is used to find the relationship between the reaction variable Y and the explanatory variables feature. The maximum training number of the model is 1000. The learning rate is 0.01, 7 level neural network is established, In the proposed model, the number of hidden nodes in the proposed model is [10, 4, 1, 10, 7, 10, 7, 4, 1], the total params is 263, as shown in Figure 17(b), and the model is established with 90% of the data and 10% of the test. The model establishment result MAE is〖1.32mm〗^2, as shown in Figure 17(c).”

 

6. Line 201 - Why, despite having 200 examples of results, only 20 were used to train the network? The fact that this number was automatically cut is not a sufficient explanation.

Thank you very much for suggestion. We have revised all the manuscript in accordance with the reviewer’s comment as follow:

The total number of samples is 200, and the training model is randomly cut at 90% (180), 10% (20) for data verification The mean absolute error (MAE) is used to evaluate the difference between the regression results and the actual results. If there is no difference between the two, the MAE is 0. As shown in Figure 8, the MAE of this model can reach .

7. Improve drawing resolution. Especially picture number 9.

Thank you very much for suggestion. We have revised all the manuscript in accordance with the reviewer’s comment as follow:

 

Figure  9. Comparison of model predictions with actual results

 

8. The article is generally good, but needs a lot of refinement to be published. Most of these corrections are simply careless and hasty drafting of the paper. Please be more accurate with the text, figures and graph descriptions for this publication.

Thank you very much for suggestion.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

No more questions.

Reviewer 2 Report

Dear Authors,

thank you for your answers and supplementing the publication with the content I suggested. In my opinion, the article is well prepared and suitable for publication.