A Technique for Bearing Fault Diagnosis Using Novel Wavelet Packet Transform-Based Signal Representation and Informative Factor LDA

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This paper proposed a method for bearing fault diagnosis using a Novel WPT-based signal representation and Informative Factor LDA.

This method uses WPT and a new rule for selecting the Mother Wavelet based on the power spectrum energy-to-entropy ratio of the reconstructed coefficients and a combination of the nodes from different WPT trees，which compensates for the shortcomings of existing bearing fault diagnosis methods that rely on the mother wavelet parameters in feature extraction.

Meanwhile, the IF-LDA feature preprocessing technology proposed in the article can provide highly sensitive features for bearing state evaluation. The effectiveness of the proposed diagnostic method in bearing fault diagnosis was verified using relevant experimental datasets.

The method proposed in this article has good innovation and reference significance for the research of fault diagnosis methods.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The reviewed article discusses the topic of "A Technique for Bearing Fault Diagnosis Using Novel Wavelet Packet Transform-Based Signal Representation and Informative Factor LDA".  In the opinion of the reviewer, the article is very interesting, although there is probably some doubt:

1) In the case of bearing diagnostics, a typical graph that gives a broad overview of the condition of the bearing is a time and frequency graph. If the time chart is characterized by an  impulsive waveform, we have a clear signal about the damage and the use of advanced techniques is not necessary. So, authors should present time and frequency plots.

2) If the tested signal shows a lot of noise and the impulse wave generated by the damaged bearing is not visible, the diagnostics must be extended with additional methods. Did the authors consider such measurement data?

3) The authors performed a reduction in sampling time. Why? Was the sampling time the same for each data set? How does time reduction affect  results?. Are there any problems with calculations for large numbers of samples?

4) The authors present a new method. Comparison with the other two is required.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

In my personal opinion, the main idea of this paper lies in using Novel Wavelet Packet Transform-Based Signal Representation and Informative Factor LDA for bearing fault diagnosis. Although some validation and identification results have been achieved, but there are still the following issues that need to be clarified and discussed:

1. The authors have demonstrated the effectiveness of the proposed method. However, it would be helpful to provide more specific details about the quantitative results, such as the specific metrics used and the comparison with other methods. 

2. The authors have provided a detailed description of the proposed method. However, it would be helpful to include more specific information about the network, such as parameters used in each stage. This would provide readers with a better understanding of the method's design.  

3. The proposed method can be essentially understood as a traditional method composing of three parts (feature extraction+feature selection+classifier). The introduction provides a clear motivation for the study. However, reviewing more literature on fault classification method will be beneficial for beginners to understand this direction. Such as, multilevel wavelet decomposition network (10.1145/3219819.3220060), multivariate singular spectrum decomposition (10.1016/j.renene.2021.02.011), and the deep order-wavelet convolutional variational autoencoder (10.1016/j.eswa.2022.119479). Besides, it would be beneficial to provide more context on the importance of fault classification method and the challenges associated with it.  

4. The authors need to further perform performance comparison to the recent works including more methods and indexes.  

5. The computation complexity and anti-noise robustnes your proposed work should be added.

6. It would be beneficial to include more detailed captions for each figure and table to help readers understand the information being presented.

7. What are the advantages and disadvantages of algorithms and how to overcome them.

These suggestions aim to provide more specific and detailed information in the manuscript, which will enhance the clarity and comprehensibility of the research. 

Comments on the Quality of English Language

It needs to be improved.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

I would like to express my gratitude to the authors for their thorough responses. The article can be published in its current form, but the authors need to perform Y-axis scaling in the FFT plots. For instance, in Fig. 2, the signal amplitude is approximately 5, while the signal amplitude in the FFT plot is 1000. The FFT plot must undergo scaling adjustments.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The author handled my comment well, and this paper can be accepted now. 

Comments on the Quality of English Language

not

Author Response

Thanks for your positive response.