Blade Crack Diagnosis Based on Blade Tip Timing and Convolution Neural Networks

Round 1

Reviewer 1 Report

Interesting paper presenting a potentially valuable application of ML to fault diagnosis. Some points of attention stand out:

 

1. 'Blades' are mentioned quite generically, apparently in the context of a rotating system, but neither the text nor the photographs and schematics of the test bed offer any insight as to the manner of loading of said blades, their geometry (which from what can be seen are flat long and thin metal strips, or their functionality (the test bed in Fig. 5 shows very thin blades that do not seem to agree with Fig. 2 - perhaps a close-up photo of the experimental setup will help clarify). It is plausible why BTT would be appropriate for such strips, which will have significant compliance to bending and thus easy to deform and produce the timing discrepancies that BTT monitors, but what is the application for such manner of blades? The authors should clarify the application, the geometry, the loading conditions and the exact failure modes.

 

2. The review of the state of the art is similarly non-specific. BBT and ML are discussed, but no mention is made to other methods (which can be argued are mre competent in certain contexts where blade stiffness is considerable, i.e. in gas turbines, compressors, pumps propellers, etc). Vibration measurements, sound measurements, potential difference measurements, etc, should all be reviewed first, in order to qualify why BBT is chosen as appropriate for the application at hand. See e.g. the following papers

 

- Krause, T., & Ostermann, J. (2020). Damage detection for wind turbine rotor blades using airborne sound. Structural control and health monitoring, 27(5), e2520.

- Abouhnik, A., & Albarbar, A. (2012). Wind turbine blades condition assessment based on vibration measurements and the level of an empirically decomposed feature. Energy Conversion and Management, 64, 606-613.

- Spitas, V., Spitas, C., & Michelis, P. (2010). A three-point electrical potential difference method for in situ monitoring of propagating mixed-mode cracks at high temperature. Measurement, 43(7), 950-959.

 

3. Fig. 1 is not easy to read or interpret. Besides the labels 'blade 1', 'Sensor' and the symbol omega, much more thorough annotation will help to understand the actual technological implementation and manner of the measurements and the layout of the sensors. Preferably Fig. 1 should be enlarged to better show the electrical networks and elements is seems to contain. A better link to Figs. 4 and 5 should be made as well. Together, they seem to convey some understanding of the layout used, but it strains the imagination of the reader.

Please consider revising and better annotating this cluster of figures (maybe, if Fig. 1 is improved, Figs. 4 and 5 will be okay as they are currently?)

 

4. The instrumentation required by BBT seems rather more complex than other vibration-/ sound- based techniques? Is this not a downside of the proposed method? And what are the advantages? Having discussed point 2 more thoroughly, the authors should elaborate on these questins, to help put their work in context and make clear its advantages and disadvantages.

 

5. It is a missed opportunity that the authors did not simulate the behaviour of blades having different parameterised cracks. It is probably too late to implement this now, but in a future study I would strongly advise to have also a predictive mechanics-based model of the system.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

The manuscript entitled “Blade crack diagnosis based on blade tip timing and 2 convolution neural network” deals with the types of blade crack faults. In this manuscript it is discussed that blade tip timing (BTT) is a non-contact vibration displacement measurement method, which has been broadly premeditated for blade vibration condition monitoring lately. Further it is worth mentioning that deep learning-based fault diagnosis methods study the logical relationships inside the data, automatically mine the characteristic laws, and this intelligently identify the fault. In recent times this research proposes a crack fault diagnostic method based on blade tip timing measurement data and convolutional neural networks for the crack fault detection of blades. In this article the author(s) claimed that this manuscript contains two main aspects: First one is the numerical analysis of rotating blade crack fault diagnosis and the second one is experimental research of rotating blade crack fault diagnosis. Further results show that the method outperforms many other traditional machine learning models in both numerical models and tests for diagnosing the depth and location of blade cracks. The findings of this study contribute to real-time online crack fault diagnosis of blades.

Over all the manuscript looks fine, contains novel idea but yet has a large amount in the way to improvement. Hence I recommend the following major revisions.

1.  Language quality needs improvement, as there are many grammatical and typos flaws in the manuscript.

2. Give a detailed motivation in the introduction section for the proposed work.

3. The authors should clearly discuss that how their presented work is the need of the time?

4. Headings of the Sections should not start with small letters.

5. Do not use bulk references in one go, for example [8-14], instead discuss all references separately?

6. Figure 3 in the manuscript is not clearly discussed, it needs more elaboration.

7. The authors should make it sure that the presented figures are not violating copyright.

8. Comparison Analysis is the most important part of any manuscript, which is not affectively discussed in this manbuscript. This issue should be resolved.

9. In Conclusion Section the authors should discuss future scope of their presented work more effectively

10. Write all references on the same pattern.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

I think the paper is very nicely written. The obtained conclusions are relevant and precise. I believe that the paper has a certain scientific contribution and that it can very serve engineers who deal with the technical diagnosis of turbomachines. For these reasons, I think that the paper will be of interest to the wider scientific and technical population.

Based on the above, I recommend the paper for publication.

Author Response

Thank you very much for your careful review of this paper. Your comments are very valuable and inspiring. We will do more in-depth research based on this paper in the future.

Reviewer 4 Report

I think it's better to add the complete closed-block diagram of the work 

Author Response

Please see the attachment

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The authors addressed all the raised points in detail. No further comments.

Reviewer 2 Report

Manuscript is revised well and so recommended for publication.