Improved Diagnostic Approach for BRB Detection and Classification in Inverter-Driven Induction Motors Employing Sparse Stacked Autoencoder (SSAE) and LightGBM

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

1. The theoretical part of this article is not very smooth to read and difficult to understand. There is room for improvement in the quality of English writing, and it is recommended to polish and revise it appropriately.

2. The position of the name annotation below Figure 3 is different from others. Please check and process such details.

3. The position of the name annotation above Table 3 is different from other tables. Please check and handle such details.

4. Please provide a supplementary introduction to the causes and difficulties in diagnosing broken rotor bars in inverter-driven induction motors in the article.

5. The clarity of the images in Figures 7 to 11 seems to be inconsistent, please verify and process them.

Comments for author File: Comments.pdf

Comments on the Quality of English Language

The theoretical part of this article is not very smooth to read and difficult to understand. There is room for improvement in the quality of English writing, and it is recommended to polish and revise it appropriately.

Author Response

Reviewer 1:

(1): The theoretical part of this article is not very smooth to read and is difficult to understand. There is room for improvement in the quality of English writing, and it is recommended to polish and revise it appropriately.

Dear Reviewer, Thank you for your feedback on our manuscript. We appreciate your constructive comments regarding the clarity and readability of the theoretical part of the article. We acknowledge that there is room for improvement in the quality of English writing, particularly in enhancing the smoothness and clarity of the theoretical exposition. Hence it is read again, and the necessary corrections are done.

(2): The position of the name annotation below Figure 3 is different from others. Please check and process such details.

Thank you for your keen observation, the caption of Figure 3 is corrected.

(3): The position of the name annotation above Table 3 is different from other tables. Please check and handle such details.

Thank you for your detailed review and for pointing out the inconsistency in the placement of the name annotation above Table 3. It is corrected now.

(4): Please provide a supplementary introduction to the causes and difficulties in diagnosing broken rotor bars in inverter-driven induction motors in the article.

A supplementary introduction about Understanding the Complexities of Diagnosing Broken Rotor Bars in Inverter-Driven Induction Motors is included in the manuscript.

(5): The clarity of the images in Figures 7 to 11 seems to be inconsistent, please verify and process them.

Thank you very much for your constructive comments regarding the clarity of the images presented in Figures 7 to 11 of our manuscript. We understand the importance of clear and high-quality visual aids in supporting the textual content and enhancing the overall comprehension of our work. Upon receiving your feedback, we carefully reviewed the mentioned figures and acknowledged that the clarity was indeed inconsistent, which could potentially hinder the understanding of the data and concepts presented.

Reviewer 2 Report

Comments and Suggestions for Authors

In this paper, the authors have investigated fault diagnosis methods for motor systems using a combination of sparse stacked autoencoders and LightGBM framework in order to achieve fault diagnosis. the paper conveyed information with sufficient detail and is somewhat innovative, but there are still a few shortcomings, as described below.

 

1.     The introduction is overly cumbersome in its description of the background and lacking description of the research methodology. The introduction should be concise, cut to the chase quickly, and focus on the innovations presented in the paper.

2.     The figure about the introduction of the dataset should not be in the introduction section, it should be adjusted to the experimental section, please make changes to this.

3.     It is necessary to refine the innovation points of the paper, make the innovation points more concise, in order to highlight the core highlights of the paper.

4.     Formulas need to be checked and interpreted for each important variable.

5.     It is suggested that the overall structure of the theoretical approach be presented in the form of a diagram to make the theoretical approach part of the paper more intuitive and to highlight the original work.

6.     This paper adopted fault diagnosis method based on sparse representation. meanwhile, dictionary learning in sparse representation is an important theoretical method. Combined with the research topic of this paper, the latest research progress of fault diagnosis method based on dictionary learning emphasizing visual and temporal information should be mentioned in introduction part.

Y. Fan, Z. Tang, J. Luo, Y. Xie, Y. Zhong and W. Gui, "Extended Shapelet Learning-Based Discriminant Dictionary for Froth Flotation Fault Recognition," in IEEE Sensors Journal, doi: 10.1109/JSEN.2024.3365706.

Comments on the Quality of English Language

The quality of the English language is decent, with no apparent lack of fluency, but the simplicity of academic expression is still lacking and needs some linguistic embellishment.

Author Response

Reviewer 2:

In this paper, the authors have investigated fault diagnosis methods for motor systems using a combination of sparse stacked autoencoders and LightGBM framework to achieve fault diagnosis. The paper conveys information with sufficient detail and is somewhat innovative, but there are still a few shortcomings, as described below.

        (1): The introduction is overly cumbersome in its description of the background and lacks a description of the research methodology. The introduction should be concise, cut to the chase quickly, and focus on the innovations presented in the paper.

     Thank you for your valuable feedback on the introduction of our manuscript. We have revised it to be more concise, directly focusing on the novelty of our research and the methodological approach. The background description has been streamlined, and we now quickly transition to highlighting our innovative contributions and the specifics of our research methodology.

          (2): The figure about the introduction of the dataset should not be in the introduction section, it should be adjusted to the experimental section, and please make changes to this.

      Thank you for your insightful feedback regarding the placement of the figure depicting the introduction of the dataset. We have duly noted your suggestion and have relocated the figure to the experimental section and also statistics of the dataset are presented in the form of a table.

         (3):  It is necessary to refine the innovation points of the paper and make the innovation points more concise, to highlight the core highlights of the paper.

       Thank you for your feedback regarding the refinement of the innovation points in our paper. We have condensed and sharpened the innovation points to highlight the core contributions of our research more effectively.

          (4):     Formulas need to be checked and interpreted for each important variable.

      Thank you for your feedback regarding the need to check and interpret the formulas for each important variable in our paper. All the formulas are checked again for every important variable and mentioned very well.

          (5): It is suggested that the overall structure of the theoretical approach be presented in the form of a diagram to make the theoretical approach part of the paper more intuitive and to highlight the original work.

      Thank you for your suggestion regarding the presentation of the theoretical approach in the form of a diagram. We agree that visual aids can greatly enhance the clarity and understanding of complex concepts. To address this, we will create a diagram illustrating the overall structure of our theoretical approach, highlighting the original contributions of our work.

        (6):  This paper adopted a fault diagnosis method based on sparse representation. Meanwhile, dictionary learning in sparse representation is an important theoretical method. Combined with the research topic of this paper, the latest research progress of fault diagnosis method based on dictionary learning emphasizing visual and temporal information should be mentioned in the introduction part.

      In this paper, we adopt a fault diagnosis method based on sparse representation, leveraging the power of dictionary learning to analyze complex signals and detect faults in induction motors. By incorporating recent research progress in fault diagnosis methods based on dictionary learning, we aim to enhance the effectiveness and robustness of our approach in identifying faults, particularly broken rotor bars, in inverter-driven induction motors.

Reviewer 3 Report

Comments and Suggestions for Authors

The reviewed article is of very good quality. The extensive comparative study deserves special attention. In terms of editing, the article is also very well prepared. However, looking at table 8, where the proposed method has 99% or more accuracy and is based on machine learning, the question arises what is the advantage over other methods? In particular, methods in the frequency domain. The frequency domain is a natural field where we analyze the state of motor damage. Authors should address this question clearly.

Author Response

Reviewer 3:

(1): The reviewed article is of very good quality. The extensive comparative study deserves special attention. In terms of editing, the article is also very well prepared. However, looking at Table 8, where the proposed method has 99% or more accuracy and is based on machine learning, the question arises what is the advantage over other methods? In particular, methods in the frequency domain. The frequency domain is a natural field where we analyze the state of motor damage. The authors should address this question.

Thank you for your positive assessment of our article and for highlighting the importance of addressing the advantage of our proposed machine learning-based method over other methods, especially those in the frequency domain, as presented in Table 8. Addressing the question regarding the advantage of the proposed machine learning-based method over other methods, particularly those in the frequency domain, is crucial for providing a comprehensive understanding of the research.