Machine Learning on Fault Diagnosis in Wind Turbines
Round 1
Reviewer 1 Report
The paper regards machine learning on fault diagnosis in wind turbine. The topic is interesting. However, there are still several questions to be solved or answered.
1. The abstract is too long.
2. In the introduction, the logic between the paragraphs doesn't seem very clear.
3. The innovations of this paper should be more clearly highlighted. In my opinion, the novelty should be focused on fault diagnosis in wind turbine rather than algorithm comparison.
4. The reasons for parameters setting of ANN should be provided. e.g. Two hidden layers are used in this study. Hidden layer 1 was set as 18 nodes and hidden layer 2 was set as 9 nodes.
5. Figs. 18-24 are not very clear.
6. The conclusions should be modified and simplified to accurately highlight this article's contributions. Future work should be added at the end of the conclusions section.
Author Response
see separate file please
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Using ANN or other Machine learning or even deep learning methodologies for fault diagnosis is an idea that has been under discussion recently, mainly due to limitations in modeling and theoretical approaches of such complicated dynamical systems. The significance of training a model based on real (not simulated) data has been also highlighted in various papers.
In general, the text is very well structured and has clearly defined topics. The abstract is a very good guide for what follows. More or less all fundamental theory details that are needed are discussed and concluding remarks are sufficient. The authors have made a concise overview of the topic and a brief reference to existing literature. They have indicated the main task of the paper among its motivation. Finally, they have pointed out the key message and the potential benefits of their work. Some comments for improvement:
1. As a general drawback, I could say that there is no reference to similar approaches (e.g. [1]) where error estimation and accuracy of machine learning methodologies have been performed on real datasets in different systems (e.g. vessels), or similar approaches with different Faults indicators have been performed on SCADA data of Wind Farms (e.g. [2]).
[1] Theodoropoulos, P.; Spandonidis, C.C.; Themelis, N.; Giordamlis, C.; Fassois, S. Evaluation of Different Deep-Learning Models for the Prediction of a Ship’s Propulsion Power. J. Mar. Sci. Eng. 2021, 9, 116. https://doi.org/10.3390/jmse9020116
[2] Encalada-Dávila, Á., Puruncajas, B., Tutivén, C., & Vidal, Y. (2021). Wind turbine main bearing fault prognosis based solely on scada data. Sensors, 21(6), 2228.
2. Justification regarding the selection of the algorithms mentioned in 2.5 should be provided. Did the authors test different models? LSTM RNN or 1D CNN could be a good fit for the problem under discussion.
3. Authors could consider shortening the 2nd section of the work. In addition, they should make sure that all material (e.g. Figure 9) included in the manuscript are both referenced and licensed as needed.
4. Authors could consider adding a flow chart or other means of visual presentation of the information provided in the last paragraph of section 2.6.
5. It would be beneficial to justify the reasoning behind the selection of the target feature used in diagnosis (section 3.2.3). Did the authors examine methodology validity and/or accuracy for different target features?
Authors should add the outcome of their research in a tabular form thus that the reader can easily see:
a. the fault rate recognition and the efficiency of each algorithm
b. false alarms per algorithm
This way easily comparisons could be made.
6. Besides the conclusions of the manuscript should be refined thus they include the main outcomes of the work, but also its impact and description of its limitations and restrictions.
7. Lastly some spelling corrections should be done, mainly due to autocorrection errors (e.g. Virus to Various)
Author Response
See attached file please
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
I recommend its publication in the journal.
Reviewer 2 Report
Authors provided an updated version of their work that could meriit publication
