Research on Fault Diagnosis of UAV Rotor Motor Bearings Based on WPT-CEEMD-CNN-LSTM

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have presented a very interesting paper that could also be important for other applications. The methodology used has been explained very well and in sufficient detail.
However, there is a lack of explanations and interpretations of the graphs shown in various sections. The authors have defined a large number of parameters for analysis, the effects of which are not easy to interpret, making it difficult to understand the graphs. 

Fig. 4: The symbols in the layers LSTM1 and LSTm2 are unclear, in particular it is not understandable why in some cases two arrows run parallel. Where can the unit described in Fig. 3 be found in Fig. 4?

Experimental Verification:
- The experimental setup shown in Fig. 5 is not helpful in this form. A 2D drawing showing the type of the entire bearing and the rotor would be necessary. 
- The bearings were obviously not preloaded, which is unusual for such an application in UAVs.
- Fig. 13 is unfortunately not explained further. The plot for 500 rpm is different from the other two. Is there an explanation for this?
- How do the authors interpret the results from Figs. 17-20?

UAV should be written out once as unmanned aerial vehicle at the first mention in the text, as should IMF as Intrinsic Mode Function. 

Does line 558 really mean Fig. 1 and not 20?

Author Response

Please refer to the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The authors present a novel approach for bearing fault diagnoses using WPT-CEEMD-CNN-LSTM. As the authors show, the signal to noise ratio is improved using their method as well as the overall accuracy and other metrics compared with other algorithms.

The article is well written and good structured. The authors sum up the fundamentals to the algorithms they use, which helps recipients to understand the content in general. The abstract is a bit confusing due to the many abbreviations, which are not explained. But in the introduction, the recipient is guided into the topic very well.

The research design is appropriate. The authors compared the performance of different algorithms on the same datasets which were provided by themselves testing artificially damages rolling element bearings in their own test rig. For further research, the performance of their algorithm to other datasets like the CWRU dataset would be interesting, so the authors can also compare their performance to algorithms of other researcher.

Some references are missing in the manuscript which I commented in the enclosed document. The authors have to clarify the references for their equations. Also, the authors have to introduce their abbreviations at first before they use them in the manuscript. After correcting this, the article can be accepted. Therefore, I recommend a minor revision.

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

 

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

no further comments.