Fault Classification of Axial and Radial Roller Bearings Using Transfer Learning through a Pretrained Convolutional Neural Network
, Kjell G. Robbersmyr 1, Tor I. Waag 2 and Thomas J. J. Meyer 2Round 1
Reviewer 1 Report
The manuscript is well organized, and the writing is clear and concise. The fault classification using transfer learning approach and other variations of machine learning algorithms is relatively new to the field of bearing fault detection. While the current method is described well in the manuscript, it is also expected that the authors could present more on the advantage of machine learning compared to traditional fault detection methods. For instance, is it truly more efficient under various conditions (different speeds, loads, different bearing failure modes, etc.)? These are the questions still remaining even after reading the manuscript. Overall, this is still a good paper for publication, and it is better if the authors can address a few comments.
1. The first sentence in Abstract seems strange, check the grammar again. Shouldn’t it be: ”Detecting bearing faults is very important to preventing...”?
2. In the experimental dataset1, explain the choice of 5kN radial load. This load seems quite small for the bearing size.
3. In the setup of dataset2, why is the speed set at 1rpm? Is there a limitation of the current method?
4. The author presented Fig 8 as an example of big challenge of fault detection in the case of high noise level. The vibrational signals in Fig 8 were in time domain, if they were transferred to frequency domain after a fft transformation, for instance, is it possible that the bearing vibrations at various characteristic frequencies can be easily determined again? The authors should take this opportunity to demonstrate the capability of current method, and show with more examples of how it outperforms the traditional methods.
Author Response
Please find enclosed the response in pdf format.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Please see the attached pdf file.
Comments for author File: Comments.pdf
Author Response
Please find enclosed the response in pdf format.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
The reviewer recommends the revised manuscript for publication in Designs.
