Rolling Bearing Health Indicator Extraction and RUL Prediction Based on Multi-Scale Convolutional Autoencoder

Round 1

Reviewer 1 Report

The paper deals with the rolling bearing failure detection method. It introduces a new algorithm to estimate a health indicator of the bearing. 

The paper is written in a pretty well manner. It has all the required components of scientific work. However, there are some things that should be changed prior to the final decision can be made. I cannot assess the adequacy of references, since they were not included. The semantics of Figure 4 is ambiguous, thus hard to comprehend. The discussion should allow assessing the actual value of the finding not only in the proposed problem solution but also in the possibility to extend it to other areas. At this moment, the scientific contribution is fairly limited. 

Author Response

I have uploaded the revised paper and reply letter to you。

Author Response File: Author Response.pdf

Reviewer 2 Report

Dear Authors,

 

I found your article very interesting, but I suggest to introduce following remarks, which have
to be added and fulfilled before publishing the paper:

 

1. Before all, I found the problem with finding the citations. I presume that the problem
   is caused by the compilation in LateX compiler. In the first paragraph I suggest to add information, what factors are influencing the bearing’s vibration level and bearing life.
   I suggest to cite 3 following papers regarding this issue:

 

108954. Ambrożkiewicz et al. (2022), The influence of the radial internal clearance
        on the dynamic response of self-aligning ball bearings, Mechanical Systems and Signal Processing 171, 108954.
108955. Zhang et al. (2021), Error evaluation of the crown profile of logarithmic generatrix roller, Journal of Physics: Conference Series, 1948, 012065.
108956. Gao et al. (2022), Experimental and theoretical approaches for determining cage motion dynamic characteristics of angular contact ball bearings considering whirling and overall skidding behaviors, Mechanical Systems and Signal Processing 168, 108704.

 

2. For better organization of the paper I suggest to prepare the list of used abbreviations
   in the paper.

 

3. In Figures 1 and 5, there are typos “dncoder”, “Offlien”,….

 

4. Figure 4 is missing.

 

5. Regarding the PRONOSTIA bearing data set, I can’t see in the paper, how long were the time-series up to failure. Moreover, the calculation time is really important for the application
   of method.

 

6. What kind of ball bearing has been used for the test. Please explain the nomenclature for bearings applied in PRONOSTIA programme. What is the difference between bearing 1_3, 2_6, 3_3, etc.? There are no information on the velocity, load or temperature. I would be also very preservative between comparing different bearings, please remember, that they can be manufactured with different quality, what strongly influence the bearing life.

 

7. In Conclusions I can’t further directions of development of your work. What I could suggest, to apply the new methodology with various health condition monitoring indicators calculated in short-time intervals.

After improving above described issues in the paper I’d like to give my positive opinion on signing my review report.

 

 

Author Response

I have uploaded the revised paper and the response letter to you.

Author Response File: Author Response.pdf

Reviewer 3 Report

After reading the submitted manuscript,my comments are as follows :

1. It seems that due to some technical glitch, references were not added in submitted manuscript.It is requested to include references and cite it at an appropriate sections.
2. In pg 3,line 92 authors mention that "  MSCAE is obtained by fusing multiple convolutional autoencoders with different convolutional kernel sizes" .How the kernel sizes were determined and what is the rationale.
3. Fig.4. is not visible.Kinldy rectify it.
4. So many spelling mistakes are observed in Fig.5,12,13 etc.Kindly rectify the mistakes.
5. In Table 1,authors mention training set and testing set.It is suggested to include ten-fold CV results since it is giving unbiased results.
6. It is recommended to include the comparison table  with already published literature (Same  PRONOSTIA experimental bench )
    to highlight the utility of proposed methodology.
7. Kindly add the limitations (if any) of the proposed methodology and future scope.

Author Response

I have uploaded the revised paper and the response letter to you.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Dear Authors,

 

I got familiar with the revised version of the manuscript, and I can give my positive recommendation. You have introduced all remarks provided by the Reviewers, which significantly increased the manuscript's scientific quality. I want to wish you all the best with Your future research.

 

Yours faithfully,

Reviewer