AI-Driven Model Prediction of Motions and Mooring Loads of a Spar Floating Wind Turbine in Waves and Wind
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThis paper introduces the method of using LSTM neural network to predict OC3-Hywind motion and mooring load for 30min. Most of the data have been well verified, the method and ideas are also clear, and the relevant deficiencies are also pointed out. However, there are still shortcomings in the introduction of theoretical knowledge and the discussion of key deficiencies, and the format needs to be improved. The followings are some minor comments that the authors are suggested to address:
1.Too many keywords and not a complete overview of the main idea of the paper. The abbreviation FOWT for Floating Offshore Wind Turbine should first be shown in the abstract. FOWT's mooring load prediction takes up most of the content in the article, but it is not reflected in the keywords.
2. In line 86, when making a general introduction to the content of the paper, it should be introduced according to the specific chapter of the paper.
3. In line 102, the description of the specific parameters of OC3-Hywind is suggested to be presented to the reader in a table directly in the paper.
4. In line 249, 30 minutes is sufficient for the prediction of all DoF motions, but 30 minutes is not sufficient for the prediction of fairlead tensions, especially in engineering practice and in the prediction of extreme sea conditions, the prediction of extreme values is crucial, which should be paid attention to.
5. In section2.2 and 5.1, it is mentioned that the simulation only considers the wave direction of 180 degrees, but this is not reasonable in the actual sea state. Therefore, the setting of the four representative cases does not take into account the factors such as normal wave direction and strong wave direction, and the significance of this paper in providing an optimization scheme for FOWT is not great.
6. The picture data is not sufficient and complete. In Figure 9, the prediction of fairlead tensions should be amplified or treated statistically where the extreme values occur. In Figure 11, the legend blocks the peaks of the three DoFs.
7. In line 395, it was mentioned that for the R2 analysis, all movements had nothing to do with Tp, but with heave R2 of Figure 10 (a) and (c), the heave R2 score of Tp 21.5s was less than 0.9, which should be explained accordingly with line 334.
8. The introduction of basic theory in this paper is lacking. In Section 4.1, the construction of the LSTM is not fully introduced, and it is recommended to use an intuitive diagram to show the relationship between the input layer and the output layer. In Section 5.2, the application of PDF and CDF is suggested to be explained by relevant theories.
9. The conclusion of the paper differs from the abstract. It is mentioned in line 425 that the proposed method is mainly applied to fatigue analysis, with the aim of abandoning the preliminary design, which is inconsistent with the gist of the abstract.
10. Some relevant literatures are suggested to be cited: 1) Optimization of mooring systems for a 10MW semisubmersible offshore wind turbines based on neural network, OE2024; 2) Study on Mooring Design of 15 MW Floating Wind Turbines in South China Sea, JMSE2024.
Comments on the Quality of English LanguageI have no comments on english language.
Author Response
See attached PDF
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThe study explains how to simulate the dynamics of a spar-type floating offshore wind turbine (FOWT) using an LSTM neural network model.
Line 249-253: The ANN model training development was discussed in order to predict a 30-minute time series for all degrees of freedom (DoF) motions and fairlead tensions. Nonetheless, more detailed quantitative metrics for different scenarios would be beneficial to mention.
The model's restrictions, particularly in relation to extreme sea states and heave prediction, should be addressed in more details.
Author Response
See attached PDF
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsThanks for addressing all my comments. I think the present version is ready for publication.
