Classification-Based Parameter Optimization Approach of the Turning Process
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe authors present a really interesting study about the classification-based parameter optimization method. This is an important topic, which is needed to be solved for the industrial applications. The manuscript is well written, the following comments help to improve its academic level:
1. Line 33: “radial or axial motion” should be written.
2. Line132: a short description should be added to the start of section 2 (3-5 lines), where the authors describe, which works will be presented here and why those.
3. Line 134: second word should be “due”.
4. Line 226: I think the normalization process should be explained more. How the data will be sorted from 0 to 1? How the maximum and minimum values are chosen?
5. Line 234: the abbreviation PCA should be explained in its first mention.
6. Line 299: second word should be “due”.
7. Line 399: in Figure 7, the blue columns are not described. Also, please add the sample size to this example as well.
8. Line 428: Please write more about these sensors. What kind of sensors are these? What is measured?
9. Line 436-440: This is truly important factors, however more explanation is necessary, why these are chosen in the study.
10. Line 453: Fourth word should be “due”.
11. Line 455: It is stated that the next steps will be turning processes, however there is drilling as well. I advise to use “machining process” instead of “turning process”. This comment also applies to other appearances of the term.
12. Line 469: Maybe “evaluated” is better here than “conducted”?
13. Line 478: The first two words “Duo to” are maybe not needed here?
14. Line 505: Fourth word should be “due”.
15. Line 538: 12th word should be “due”.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThe paper titled “Classification-based parameter optimization approach of turning process” offers valuable insights into the optimization of turning processes and classification models. The manuscript's scope is well-defined and comprehensible. However, some minor revisions, mainly in the organization and structure, are recommended before publication:
Keywords:
- Consider adding keywords related with applied models or algorithms.
Related work:
- Section 2 provides a review of parameter optimization and pattern recognition methods, which should be integrated into the introduction section for better coherence and flow.
- Line 134: Correct “duo to” to “due to”.
The Proposed Method:
- In the Data Pre-processing subsection: Could you provide more detailed information regarding the types of sensors used? For example, specifying the use of dynamometers for force measurement, accelerometers for vibration detection, displacement sensors (such as LVDTs), thermocouples and infrared sensors for temperature monitoring, torque sensors for rotational force, acoustic emission sensors for detecting microcracks, electric current sensors for energy consumption, and optical sensors for precision measurement.
- Related to the previous point: Could you clarify the type of sensors referenced in line 428? Identifying the exact sensors used would enhance clarity and understanding.
Experimental case study:
- Lines 453, 478: Correct “duo to” to “due to”.
- Line 731: Clarify how this statement impacts time reduction or productivity improvement: "Each product has 4 axes that need to be treated".
Conclusions:
- Line 761: Change the comma after "process" to a period.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsAiming at the turning process, a classification-based parameter optimization strategy with three steps is developed to improve the productivity in this manuscript. However, some issues need to be solved to improve the manuscript.
1. Some formatting issues need to be noted. (1) The symbol explanation paragraph after the Equation should be written without spaces, and the first letter of "where" should be lowercase. (2) The font size in some figures is too small. (3) When explaining the steps of a certain process, it is better to replace the numbers 1, …, 3 with the form of Step 1, ..., Step 3. (4) The main shaft, workpiece, tool, and sensors should be marked clearly to show their configuration status in Figure 8. The XYZ coordinate axes should also be marked.
2. It is necessary to explain the meaning of each equation and even the logical relationship between them, rather than simply listing the equations.
3. Turret vibration in x-axis is missing in Table 1. It seems that cutting parameters (such as rotation speed of the spindle, depth of cut, and feed rate) should be considered as parameters to be optimized rather than feature parameters as evaluation indicators.
4. It would be better to express the sampling frequency in “Hz” rather than “s-1”. The conversion of sampling frequency from 4000 Hz to 10 Hz can be achieved by resampling. Why are their effective values calculated? More importantly, the sampling frequency of 10 Hz is too low, which causes all effective information of the signal above 5 Hz to be lost. At the same time, in data preprocessing, the original signal is scaled to a zero-mean signal, which causes the bias component of the original signal (i.e. the average value of the original signal) to be lost. Vibration signals between a-5 Hz (a is an infinitesimal value greater than 0) are usually not caused by cutting whose rotation speed is larger than 300 r/min. It is doubtful that different cutting stages can be identified accurately under such conditions.
5. What is the mean of “rough external cylindrical cutting and drilling”? Please explain the difference between 5th and 6th stages. It is said that “each experiment in this search is repeated three times, and then these results are averaged”. What do “these results” mean? It seems unreasonable to average either the directly measured vibrations or the evaluation results.
6. Please explain the number of experiments (i.e. the data size) performed to construct the training and test sets and the main cutting parameters used in the experiments.
7. The sampling frequency is 10 Hz, and the window length is 40, i.e. 4 seconds. There will be data spanning two different stages. How to classify them?
8. How to identify tool wear status in Figure 14? The vibration of the cutting system is not only related to tool wear, but also to tool specifications, cutting parameters, etc. Whether the method of judging the tool wear status by comparing the vibration curve is reasonable. It is more accurate to judge the wear status by the physical wear size of the tool.
9. The proposed method seems to only determine which stage can be optimized. The optimization parameter needs to be selected through further experiments, while the method itself cannot provide the optimization result. If so, the function of the developed optimization method is not amazing enough.
10. According to Section 4.6.2, parameter optimization is performed based on the vibration threshold and the lower envelope curve of the vibration. In this case, the result of parameter optimization should be the reduction of vibration. In addition, smaller cutting parameters (such as lower depth of cut, lower rotation speed, and lower feed rate) may also meet the vibration threshold requirements. What is the constraint? How to ensure that the optimized cutting parameters can improve the productivity?
Comments on the Quality of English LanguageAfter revision, proof reading is needed.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 4 Report
Comments and Suggestions for AuthorsGeneral comment: The research topic is very current; the manuscript is interesting and may be of interest to future readers. The authors have made a great effort to comprehensively conduct and present the research. I propose that the paper be published when the following minor changes and additions are made to the manuscript.
The title is well-worded.
In the abstract, it should be stated what the goal of the research is.
Keywords should be expanded, with some words such as: cutting tool, CNC, etc.
In the text itself, it is necessary to separate units from numbers.
The proposal is to cites some equations in the paper.
The Figures' quality is satisfactory.
The introduction very well introduces the reader to the problem under investigation. I suggest to better present the scientific justification of the research and to add the following very important statements:
- “In order to improve the machining process and reduce manufacturing time and costs, the industry is increasingly turning to modeling techniques.” (Consult and add the following research: https://doi.org/10.3390/app14145980)
- “Turning makes 40% of the total processing, and the number of lathes makes 25-35% of the total machines in the workshops.” (Consult and add the following research ttps://doi.org/10.18485/aeletters.2022.7.4.2)
For a clearer work methodology, a more detailed description of the equipment used should be presented.
It is unclear how and with what equipment the measurement was made: speed parameters, vibration parameters?
The text related to Figure 8 should be better described and connected with the figure itself.
Does Figure 8c show a representative sample that can be discussed in the research. Because it is a very complex working part of the "cross shaft".
The presentation of research results and their explanation are very well and clearly described.
The conclusion is clear. The proposal is to mention the main scientific justification of the conducted research.
The choice of references is good.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 3 Report
Comments and Suggestions for AuthorsAiming at the turning process, a classification-based parameter optimization strategy with three steps is developed to improve the productivity in this manuscript. According to the comments, the manuscript has been revised. But not all the comments have been well addressed. Revision corresponding to the comments below should also be done in the revised manuscript.
1. The conversion of sampling frequency from 4000 Hz to 10 Hz can be achieved by resampling. Why are their effective values calculated? More importantly, the sampling frequency of 10 Hz is too low, which causes all effective information of the signal above 5 Hz to be lost. At the same time, in data preprocessing, the original signal is scaled to a zero-mean signal, which causes the bias component of the original signal (i.e. the average value of the original signal) to be lost. Vibration signals between a-5 Hz (a is an infinitesimal value greater than 0) are usually not caused by cutting whose rotation speed is larger than 300 r/min. It is doubtful that different cutting stages can be identified accurately under such conditions.
2. The vibration of the cutting system is not only related to tool wear, but also to tool specifications, cutting parameters, etc. Whether the method of judging the tool wear status by comparing the vibration curve is reasonable. It is more accurate to judge the wear status by the physical wear size of the tool.
3. The proposed method seems to only determine which stage can be optimized. The optimization parameter needs to be selected through further experiments, while the method itself cannot provide the optimization result. Please explain in detail how to optimize the parameters.
4. According to Section 4.6.2, parameter optimization is performed based on the vibration threshold and the lower envelope curve of the vibration. In this case, the result of parameter optimization should be the reduction of vibration. In addition, smaller cutting parameters (such as lower depth of cut, lower rotation speed, and lower feed rate) may also meet the vibration threshold requirements. What is the constraint? How to ensure that the optimized cutting parameters can improve the productivity?
Author Response
Please see the attachment.
Author Response File: Author Response.pdf