An Automatic Data Augmentation Method for Working Condition Diagnosis of Rod Pumping Systems Based on Teacher Knowledge
Round 1
Reviewer 1 Report
The manuscript under the title "An automatic data augmentation method for working condition diagnosis based on teacher knowledge" discusses the automatic data augmentation method. This manuscript requires significant modifications, including using technical terms such as what is meant by teacher knowledge. Besides, the structure of this manuscript is very complicated, and cannot be easily understood by a reader. Based on the quality and provided results, I am not convinced of the current manuscript's acceptance.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This paper presents a study on the " An automatic data augmentation method for working condition diagnosis based on teacher knowledge". This is an interesting manuscript for this journal but I suggest a major revision. Here are some bugs in this article to help the authors to profit from this article, but if the authors can't do these comments (point by point) the article will be rejected.
================================
1) General comments:
1a) Some grammatical error sees in the article. Please take time to improve the language.
2b) Article is too long (the same as book chapter). The authors must decrease.
================================
2) Keywords and Highlights:
2a) The authors must update their old keywords.
2b) The authors must add highlights in their article.
================================
3) Abstract:
3a) The abstract doesn’t have novelty in it. The authors should rewrite the abstract with main novelty in it.
3b) What is the main purpose of the article? The authors should focus on novelty on this section. Please highlight it.
================================
4) Introduction and Literature Review:
4a) The introduction is very long. The authors should be eliminating some part of it. And also take part Introduction and Literature Review
4b) The authors must proper one table and compare their articles to literature review (major comment).
================================
5) Methodology:
5a) I strongly suggest authors add the statical parameter for data collection for this article (major comment).
================================
6) Results and Discussions:
6a) Some figures are not high resolution in the article (figure 7 and 8). Change with high resolution.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Hi Authors
Thanks for submission.
TITLE
An automatic data augmentation method for working condition diagnosis of ???? based on teacher knowledge - Kindly review
ABSTRACT
Seems quite subjective. Can you pls make it quantitative? By adding sone quantitative results or its applications findings?
Keywords> Quite lengthy - Kindly revisit
INTRIODUCTION
Reference entering Pls use MDPI format [1]
The headings are not as per MDPI format kindly update?
For the working condition diagnosis of the rod pumping system, multi-category faults like ???? Can you list some in a table?
of failure types, and the model classification effect is not good. In order to prevent {Pls add reference]
Research motivation, research gap, research novelty are missing in introduction.
This paper proposes? Pls revisit Can be The research propose?
How your objectives 1, 2, 3 can be in present tense please revisit?
The rest of the paper is structured as follows: In the second section, the related algo- 116 rithms for automatic data augmentation and teacher knowledge are introduced to facili- 117 tate the understanding of the entire algorithm model. In the third section, a teacher- 118 knowledge-based automatic data augmentation model for operating condition diagnosis 119 is proposed, and parameters and structural details are introduced. The fourth section ap- 120 plies this method to the experiment of the actual indicator diagrams data set, and explains 121 the experimental design, experimental results, and comparative experimental analysis. In 122 the fifth section, the advantages and disadvantages of this method and future prospects 123 are discussed. The sixth section is the conclusion part-----
Lines 116 to 124 pls follow MDPI format Its not thesis summary of the chapters?
In introduction, please enter some developed models either in form of table or in form of figure than discuss about their shortcomings?
Introduction in current form is very weak please improve it.
RELATED ALGORITHMS
Is it part of introduction? or literature review? So place it before your objectives in Introduction section.
Figure 1,2 Reference??
Equations 1, 2, 3 Reference ??? You can use a table for describing these equations with references?
PROPOSED METHOD
problem of working condition diagnosis like ???, which involves parameter optimization like ???// and image processing methods like ??? Pls explain????
Suppose ?? represents an image sampled from dataset X, and ???? represents an augmentation function with parameter Ï•. In methodology you can not use supposing values? Pls be specific?
This objective is solved by alternately updating Ï• and θ. However, adversarial data augmentation meth- 248 ods cannot improve model generalization well without regularization or restrictions on 249 the size of the search space. Because maximizing Ï• tends to ignore the intrinsic infor- 250 mation of the image ??. Therefore, instead of regularization based on prior knowledge, we 251 use the teacher model. ????? Pls discuss models inputs and differentiate between different variables? Further also specify model limitations and any boundary conditions? ???
3.2. Automatic augmentation method based on teacher knowledge - Pls elaborate and rewrite to make it more clear????
Figure 3 pls differentiate clearly between your and traditional approach. Figure 3 make it more readable?
The function objective is set as follows Reference ????
For the target model ????? Not clear which Target Model Authors are describing?, this target has the same components as the adversarial data augmentation method? How are Authors comparing model with method Pls revisit????? But this augmentation function not only requires maximizing the loss of the objective function ????? How elaboration is required????,
To improve understanding kindly label Figures 3 and 4 properly???
Figure 4 student model? what is this?
Figures 3 and 4 References ????
Figure 5 is not readable? The x and y axis values are not visible? what these values are pls add axis and label?
How equation 5 wass derived not clear?
Equation 6 citation missing? Have you taken these equations from some references? Pls make the text clear?
Pls renumber equations 6 to 10? Why these equations there should be some rationale?
EXPERIMENTS
Up to line 353 out of total lines i.e. 553 in paper the Authors have not yet described any thing they have carried out???? The Authors kind attention is requested.
Collect indicator diagrams data points from actual pumping units ? ?? Pls share some sample??
The data is preprocessed here.???? Where and which data?/
What are x and y in Figure 6 what are their axis lables and units??? Reference of Figure 6 and Equation 11 is also desired???
Sample selection not clear? Any reference from literature in this regard?
There are a total of 700 samples of each type, and a 369 total of 4200 indicator diagrams. The data augmentation method is used to uniformly in- crease the number of samples to 1,000 per class, so that a total of 6,000 samples are obtained.??? in figure OR table PLS SHOW SOME TYPICAL COMBINATIONS OF samples, indicator diagrams, class ? to make the experimentation more clear?
What is Cifar10 for readers make it clear??
Why 8:2 and 800, 200 training testing sets pls develop rationale????
For the problem ??? Yu have not described yet?
Process of experiment is also all literature review? How you claim it under your experimentation? Pls revisit?
4.3. Experiment results and analysis Line 402 very unclear and not linked with what you have described above? Pls rewrite and organize it properly for the readers for better understanding?
Figures 7 and 8 can you compare it with some existing trends described in literature?
Can you pls enter titles of axis along x and y axis?
How Table 1 parameters were determined pls make it clear???
DISCUSSIONS
Should be based on your novel findings currently these are having very narrow relevance with your scope, keywords and experimentation?
CONCLUSIONS
Should be from your objectives? Currently these are very wake?
REFERENCES
Pls add few from MDPI Journals also.
A very detailed reworking is desired as per the MDPI format and guidelines to bring the scattered material in line for possible consideration of publication? As it is not a review paper, Authors are requested to focus on their own work currently which seems to be very shallow.
Authors are requested to rework based on above guidelines to make it more understandable and meaning full for the readers.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
The author answers all changes inn the article.
Author Response
Thank you very much for your comments!
Reviewer 3 Report
Hi Authors
Thanks for the submission of the revised version to MDPI.
2. Related algorithms? Does it fall under introduction or materials methods? Pls assess.
Table 1 - What do you mean by Our method?
Column 1 of Table Please replace caption Author with Reference? Pls add reference no. as per MDPI format also.
Pls expand more columns in Table 1. How accuracy was determined please comment?
Figure 1 Reference please
Table 2 How enhanced data achieved? Pls describe?
Figures 7 and 8 what is traditional augmentation pls explain? Is it by one method or by combination of methods?
Table 3 why these two methods comparisons with your work? Kindly discuss rationale?
The discussion part needs special attention. Kindly expand this part. In current form it is very weak.
There are lot of clues which you missed/did not address in our earlier comments for the improvement of this section. Kindly refer to our previous comments?
The data, sampling, data in Figures and Discussion part need more in depth review pls by the Authors.
Various interesting workds used in conclusio9ns but did not find those words in paper? Kindly review. i.e. Intelligent oilfields, cifar10, dynamometer diagrams, optimal augmentation strategies, operating condition diagnosis?
References in reference section are not as per the MDPI format.
Thanks
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 3
Reviewer 3 Report
Hi Authors
Thanks for submitting the revised version.
I appreciate the perseverance and efforts authors made for the improvement of the manuscript.
Few minor corrections/suggestions are attached below:
(1) We propose an automatic data augmentation method based on teacher knowledge to solve the problem of 90 insufficient samples for the working condition diagnosis of rod pumping systems. With 91 the teacher model, the transformed images can avoid losing their inherent meaning and 92 also make the adversarial enhancement more informative without the need to manually 93 adjust the parameters. (2) In this paper, a standard indicator diagrams data set is con- 94 structed according to the "cifar10" data set format for the application of the working con- 95 dition diagnosis model. This data set construction method can be used for actual oilfield 96 database management to provide support for subsequent analysis and processing. (3) A 97 Sustainability 2022, 14, x FOR PEER REVIEW 3 of 18 neural network is proposed for data augmentation of indicator diagrams, which can per- 98 form gradient descent to update parameters and simplify the design of the search space. ====== Pls do not use Sr. No. in paragraphs.
The rest of the paper is structured as follows: In the second section, the related algo- 100 rithms for automatic data augmentation and teacher knowledge are introduced to facili- 101 tate the understanding of the entire algorithm model. In the third section, a teacher- 102 knowledge-based automatic data augmentation model for operating condition diagnosis 103 is proposed, and parameters and structural details are introduced. The fourth section ap- 104 plies this method to the experiment of the actual indicator diagrams data set, and explains 105 the experimental design, experimental results, and comparative experimental analysis. In 106 the fifth section, the advantages and disadvantages of this method and future prospects 107 are discussed. The sixth section is the conclusion part. ======== What about first section?
Line 109 Materials and Methods Pls correct
Table 1 Pls replace us with Method Proposed in this study
Line 173 , AutoAugment Check typo
Table 2 Pls add "description" in tables 2 & 3 heading for column 1
Line 463 The loss curve and accuracy curve are shown in Figure 7 and Figure 8. ====== Figures 7 & 8.
Conclusion ===== Conclusions Typo Line 535
References need to be as per the MDPI format.
Pls modify all tables and figures captions statements. Make it more appealing removing the repetition words.
All the Best
Author Response
Thank you very much for your advice and help! Please see the attachment for the modification results.
Author Response File: Author Response.pdf
