A Machine-Learning-Based Method for Ship Propulsion Power Prediction in Ice

Round 1

Reviewer 1 Report

In many places the scientific and technical topics are not well, nor exhaustively described.

The relevant formulas are badly described, in some parts some further mathematical details must be added, e. g. some details of the different fuel consumption models must be added, in order to comprehend the different approaches considered, and to explain how the adopted machine learning algorithms simplify data processing and computations.

Mathematical notation is poor, and some formulas are not well written or explained in the text. Some examples (not exhaustive list):

 * Eq. 3, Pag. 14: explain wy no normalization constant is present;

 * Eq. 5, Pag. 14: a typo is present in the formula, it seems that the sigma symbol is missing;

 * Eq. 6, Pag. 14: poor mathematical notation, also in the following explaining text;

 * Matrix W=[…] Pag. 16: disordered set of numbers, whose computation is not correctly explained in the text.

Many figures have to be improved:

 * Figs. 1, 2, 10, 11, 12: very poor quality;

 * Fig. 3: no units are reported (if non-dimensional quantities are reported, this must be explained in the text ant in the respective figure caption).

* Figs. 5, 6: not so much informative.

Some citations to References are wrong, citing author names instead of Ref. numbers […].

Excessive (and sometimes not proper) use of acronyms, someone of them not defined. The presence of the “Nomenclature” table at the beginning of the manuscript is not enough, each acronym must be also defined the first time it appears in the text. Moreover the excessive use of acronyms results in a not easy reading.

The analysis of the results is very much not clear and ineffective. Consequently, also the conclusions are very poor. The real scientific value of the work performed and of the achieved results does not clearly comes out. Moreover also the included figures are not so helpful.

The English is very poor, and must be strongly revised. Most of the manuscript is not clearly comprehensible. In some cases not proper words are used. The manuscript looks more like a first version of notes to be further developed and improved in order to be transformed into a more presentable document, it does not appears as sound finished scientific article. Moreover, no careful proofreading seems to have been done.

Author Response

Thank you for your review. The response can be seen in attachement.

Best regards

Author Response File: Author Response.pdf

Reviewer 2 Report

The manuscript entitled " A machine learning-based method for ship propulsion power prediction in ice" is interesting to discuss. Artificial Neural Network is used to predict the ship propulsion power. The authors make improvements to the method. This gives the good results and small error obtained from the ANN. Some parts need to be commented to make the explanation's author clear. Here are the comments:

1. The abstract is not written clear especially since reading in the first sentence as follow: "Accurate prediction of propulsion power plays an important role in ensuring the safety of ship navigation when sailing in ice regions". Author should explain clearly why ice regions are important here. This should be explained in the next sentence.

2. In Introduction, author did not explain the gaps between reference papers. The authors should explain what is the lack, implicitly, on a paper and this should be covered in the paper. Recently, the authors just explain what people did but they did not dig more detail. 

3. In section of ANN model, people with less knowledge about this topic will confuse what this paper presents. Brief explanation of what the ANN model should exist. It should also present some equations that are used in the paper. 

4. In Figure 4, the R is 0.69 which is lowest accuray that others data. Can this data be improved? Please make justification if you can do or not. 

5. Figure 5 and 6 are good for presenting the normalization data. But the reviewer would like ask how the value generated. Because the authors use the solid line, the values also represent for small increment. The reviewer did not think so. Maybe it is better to present in dot. Please comment this. 

6. The conclusion is good enough but it can be more improved.

Author Response

Thank you for your review. The response can be seen in attachement.

Best regards

Author Response File: Author Response.pdf

Reviewer 3 Report

The authors propose the use of Artificial Neural Networks (ANN) for predicting the propulsion power of polar ships. Although the methodology itself is not novel, applying ANN to this specific application is interesting. However, there are several concerns that need to be addressed to improve the quality and readability of the paper.

 

1- Handling Randomness in ANN Optimization:

The authors employed Particle Swarm Optimization (PSO) to optimize the ANN. Given that ANNs have random weights, it is important for the authors to address how they handled the issue of obtaining different responses in each run. If this issue persists, it should be clarified as a limitation in the conclusion, potentially suggesting further research in this area.

 

2- Why not Deep ANN?

Considering the advancements in Deep Learning, it would be valuable for the authors to discuss why they chose to use traditional ANN instead of more advanced architectures such as Deep ANN. This discussion will provide insights into the decision-making process and potential avenues for future research.

 

3- Bridging Traditional Models and ANN:

To provide context and establish the relevance of ANN in this study, the authors should bridge the gap between traditional machine learning models and ANN. They can include a statement in the introduction that highlights the limitations of traditional models in dealing with large datasets with significant variations, and then emphasize that ANN, including Deep ANN, is a promising alternative. Please provide following citations to support this statement [x, xx].

[x] https://doi.org/10.1007/s12145-022-00885-6 [xx] https://doi.org/10.3390/fishes7060385

 

4- Contribution of the Present Study:

In the last part of the introduction, it would be beneficial for the authors to summarize the contributions of their study using bullet points. This will clearly outline the unique aspects and key findings of the research.

 

5- Explanation of Datasets:

The paper lacks a comprehensive explanation of the datasets used in the study. It is crucial to provide details about the datasets, including their sources, sizes, and relevant characteristics. This information will help readers understand the data used for training and evaluating the ANN model.

 

6- Model Illustration and Methodology:

To enhance clarity, the authors should present a model illustration at the beginning of the methodology section. This illustration should provide a visual overview of the methodology and its steps. Additionally, the methodology section needs to be rewritten coherently, with a consistent level of detail and logical flow. Ensure that each step is explained thoroughly, enabling readers to easily follow the process.

 

7- Evaluation Criteria:

In the methodology section, it is essential to provide the evaluation criteria used for assessing the performance of the model. Please include relevant formulas or equations to ensure transparency and replicability.

10- Additionally, it would be valuable to include a comparison with traditional models like Random Forest. This comparison should consider metrics such as accuracy and computation time to highlight the advantages and disadvantages of using ANN in this application.

9- Conclusion and Discussion:

The conclusion and discussion section requires improvement. The authors should expand on the research outcomes and implications, providing deeper insights and potentially suggesting future research directions. This will enhance the overall impact and significance of the study.

Author Response

Thank you for your review. The response can be seen in attachement.

Best regards

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

The authors well addressed the comments, and the manuscript is acceptable.

Author Response

Please see the attachment

Author Response File: Author Response.pdf