A Novel Method for Gas Turbine Condition Monitoring Based on KPCA and Analysis of Statistics T2 and SPE
Round 1
Reviewer 1 Report
Dear Editor, Dear Authors,
I read the paper titled “A novelty method of tested data processing for gas turbine condition monitoring”.
The topic of this manuscript partially falls within the scope of Processes. In my opinion there are other journals of MDPI that match better the topic of the paper
English must be revised by a native speaker. There are a lot of typos and sentences not well structured.
These are my comments and remarks:
· Title. Title is not so clear in my opinion. “Novelty” is a noun and could not be used as adjective of the word “method”. “Tested data” is not informative, tested on what? The two features used (T2 and SPE) are not mentioned.
· Line 31. “Method based on knowledge”. It is not clear the meaning. The other methods are not based on the knowledge? Specify better and add references.
· Introduction. Physical modelling is missing in the introduction. Based on the specific component, there are several physical-based models in literature. It is not clear which type of fault the authors are interested on. Or which component they focus on. The introduction must be revised deeply.
· All the manuscript. Avoid sentences like “Paper proposes”. Which paper? The one of the authors? It is more clear use “This paper proposes”.
· Line 51. The authors introduce T2 and SPE statistics and suppose the reader already know them. It is better to describe them a little bit, just explaining the acronyms.
· It is not clear the passage from equation 2 to 3. Why does lambda include the product of phi and vi? In Eq. 2 was simply a product.
· Equation 5. Check the last equality, it is not easy to follow and I don’t understand why alpha_k is outside the summation k-indexed.
· Equation 8. What is alpha_i^k?
· Line 93. “Kernel function strengthens”, but the kernel is not explicitly defined. Just say that K(x_i,x) is defined as Kernel just after eq. 8.
· Line 98. “To achieve the fault detection of gas components”. Gas components is wrong. It is “gas turbine components”.
· Equation 11. Specify what “Lambda” is?
· Equation 12. Specify what does the subscript “th” mean. Specify what “F_alpha” is?
· Equation 15. Probably sigma^2 at the denominator of the last equality is missing.
· Equation 16. the third equality must be detailed better. Since it is a partial derivation with respect to v_k, it is expected to see the dependencies with v_k (as in the last equality).
· Line 153. “It is widely known”. It is better to add references to support this statement.
· Line 155. Could be “rotor” instead “rotator”? Just a question, I’m not sure.
· Figures 1 and 2. Add the unit of measure to the y-axis of all the figures.
· Line 168. “How pressure” instead of low (or high) pressure.
· Results. Describe the type of damage founded on the real turbine.
· Line 188. The threshold. Detail how the threshold is defined.
· Line 199. “gas path components”. It is better to use “gas turbine components”.
· Conclusions. The conclusions are too short. Extend this part.
The authors propose a procedure that experimentally proves to be successful. For that reason, I think it is worth publishing it. But not in the present form. The paper must be revised, from a technical point of view and also from a language point of view. Based on the comments above, I suggest a major revision of the paper.
Regards
Author Response
Dear reviewer:
Modifications have been made in response to your suggestions for amendments and the English expressions throughout this paper have been revised.
Author Response File: 
Author Response.docx
Reviewer 2 Report
1) The novelty is unclear, as the formulae from (1)-(23) are quite standard. Not sure which part is your original contribution.
2) A algorithm should be provided by the end of Section 2.
3) Introduction should be enhanced by referred to mostly recent special issues on fault diagnosis and conditional monitoring (e.g., 10.1109/TIE.2015.2417511, and https://doi.org/10.1016/j.renene.2017.10.059)
4) Robustness against noises and bad data should be better discussed and explained.
Author Response
Dear reviewer:
Modifications have been made in response to your suggestions for amendments
Author Response File: Author Response.docx
Round 2
Reviewer 2 Report
The manuscript has been revised and improved. The contribution of the paper should be a fault diagnosis algorithm application. No further comments, and the paper is ready to be accepted.
