Optimization of Structures and Composite Materials: A Brief Review

Round 1

Reviewer 1 Report (Previous Reviewer 2)

Comments and Suggestions for Authors

The article is devoted to Optimization of structures and composite materials: a brief

review. Neural networks trained to predict loads are crucial for estimating the loads on various aircraft components in different flight scenarios. In addition, machine learning facilitates topological optimization by identifying patterns and optimizing the distribution of materials. This study examines these applications, highlighting the scientific and practical contributions of machine learning to aeronautics.

However, there are some comments on the work:

1. The Abstract section needs to be rewritten, reflecting the relevance of the problem being solved and the scientific novelty of the solution obtained.

2. The keywords need to be corrected by adding special terms characterizing the study.

3. The introduction section should indicate the novelty of the study. At the end of the introduction section, it is necessary to define the purpose of the scientific study.

4. There are certain disadvantages of combining neural networks (NN) and genetic algorithms (GA) for structural optimization: Redundancy of NN structure encoding. Heterogeneity of binary string. Difficulties in searching in continuous spaces of high dimensionality. How are these issues currently addressed?

5. When optimizing composite materials, it is necessary to predict the loads acting on the structures in which these materials are used in order to understand their ability to withstand forces and to better optimize them. Which sections of artificial intelligence and algorithms do the authors consider the most effective?

6. How can the reliability of composite materials of aircraft be predicted?

7. The list of cited sources should include more modern publications on predicting the reliability of composite materials.

8. More quantitative characteristics of the assessment of the effectiveness and adequacy of models using machine learning and artificial intelligence should be provided for the articles included in the review.

9. The conclusions should be structured, highlighting the main scientific and especially practical results obtained, as well as recommendations for designers and mechanical engineers.

Author Response

The article is devoted to Optimization of structures and composite materials: a brief review. Neural networks trained to predict loads are crucial for estimating the loads on various aircraft components in different flight scenarios. In addition, machine learning facilitates topological optimization by identifying patterns and optimizing the distribution of materials. This study examines these applications, highlighting the scientific and practical contributions of machine learning to aeronautics.

However, there are some comments on the work:

1. The Abstract section needs to be rewritten, reflecting the relevance of the problem being solved and the scientific novelty of the solution obtained.

- Thank you for your constructive comment. The abstract was rewritten (letters in green). The main problem to be solved in this study is to look for solutions to accelerate the process of optimizing geometries and compositions of composite structures in relation to traditional design of experiment (DOE) methods, or using numerical simulation which, depending on the complexity of the model, can take considerable time to simulate each point in the multivariable space defining the composite structure (structure dimensions, matrix/reinforcement composition, reinforcement architecture and volume fractions, etc).

2. The keywords need to be corrected by adding special terms characterizing the study.

- Thank you for your constructive comment. We added Review to the keywords list (limited to 6 keywords)

3. The introduction section should indicate the novelty of the study. At the end of the introduction section, it is necessary to define the purpose of the scientific study.

- Thank you for your constructive comment. The introduction was rewritten (letters in green). The same comments were included in the introduction section.

4. There are certain disadvantages of combining neural networks (NN) and genetic algorithms (GA) for structural optimization: Redundancy of NN structure encoding. Heterogeneity of binary string. Difficulties in searching in continuous spaces of high dimensionality. How are these issues currently addressed?

- Thank you for your constructive comment. These disadvantages and limitations of combing NN and GA were highlighted (letters in green).

5. When optimizing composite materials, it is necessary to predict the loads acting on the structures in which these materials are used in order to understand their ability to withstand forces and to better optimize them. Which sections of artificial intelligence and algorithms do the authors consider the most effective?

- Thank you for your constructive comment. In aeronautical engineering, loads can be categorized as resulting from manoeuvres (on the ground and in the air), discrete gusts, and turbulence. For manoeuvres, aeronautical regulations already specify the most critical cases that the structure must withstand. In the case of discrete gusts, the regulations define a gust profile (1-cosine) with intensities that vary with the flight condition as well as the length of the gust, and the aircraft structure must endure the worst-case scenario. On the other hand, turbulence involves a spectrum of loads, with regulations requiring the use of the von Karman PSD. Given this context, it was deemed unnecessary in the present work to use artificial intelligence to predict aeronautical loads, as these are well defined by aeronautical regulations. 

6. How can the reliability of composite materials of aircraft be predicted?

- Thank you for your constructive comment. A paragraph was introduced.

7. The list of cited sources should include more modern publications on predicting the reliability of composite materials.

- Further references were introduced regarding reliability prediction of composite materials and results from CDM simulation methods used to train NN’s, also answering to the previous point.

8. More quantitative characteristics of the assessment of the effectiveness and adequacy of models using machine learning and artificial intelligence should be provided for the articles included in the review.

- Thank you for your constructive comment. A comment regarding these quantitative characteristics, such as precision and accuracy, compared to experimental and simulation methods, was introduced (letters in green) in Structural Optimization section.

9. The conclusions should be structured, highlighting the main scientific and especially practical results obtained, as well as recommendations for designers and mechanical engineers.

- Thank you for your constructive comment. Conclusions were structured to address reviewer comments (letters in green).

Reviewer 2 Report (Previous Reviewer 1)

Comments and Suggestions for Authors

The authors addressed my previous concerns and I think the paper looks good now.

Author Response

The authors addressed my previous concerns and I think the paper looks good now.

 

- Thank you for your revision

Reviewer 3 Report (New Reviewer)

Comments and Suggestions for Authors

This paper provides a brief review of machine learning (ML) applications in aircraft structures and composite materials. While this is an interesting topic and of great importance, the presentation is not ideal. Therefore, a major revision is necessary to be reconsidered for publication.

1.      Please use the word template provided by the journal website (https://www.mdpi.com/files/word-templates/eng-template.dot).

2.      Why are some texts highlighted in yellow or red?

3.      The article title in the manuscript is different from the article information.

4.      Page 4 line 100, the authors claimed that “General bibliometric research on the optimization of composite materials reveals a growing trend in research activity”. However, it is hard to tell from Figure 5 since there were less than 10 documents per year in the last 8 years. I believe that it is an upward trend, but your search keyword of “aircraft” might filter out many relevant articles.

5.      The abbreviation of machine learning (ML) is defined twice (line 121 and line 165). Same issue for neural networks (NNs).

6.      On page 7 line 162, the statement “DNNs are a specific type of NNs, however, DNNs have more hidden layers compared to NNs” is misleading. They are not comparable since DNN is a sublet of NN. In fact, it is debatable whether to have a separate section for DNN as some references before Section 3.1 use DNN with more than 3 hidden layers, for example, [20]. Most modern NN applications are deep based on this definition.

7.      On page 10 line 288, “Optimizing an aircraft is extremely important, as this is a fundamental practice for improving it”. Do you repeat the same statement twice?

8.      Also on page 10, the subsection number should start with 5.1.

9.     The section title of section 5 (Survey of Previous Literature) can be confusing as the entire paper is about the previous work.  

 

10.   It is not necessary to have Sections 6 and 7. I would recommend moving those contents to Sections 5 and 8.

11.   Further language polishing is recommended.

Author Response

This paper provides a brief review of machine learning (ML) applications in aircraft structures and composite materials. While this is an interesting topic and of great importance, the presentation is not ideal. Therefore, a major revision is necessary to be reconsidered for publication.

 

1. Please use the word template provided by the journal website (https://www.mdpi.com/files/word-templates/eng-template.dot).

- We used the optional latex template

2. Why are some texts highlighted in yellow or red?

- This is a revised version. The highlighted text in different colours was in accordance with the first response to reviewers, and submission of this current version. 

3. The article title in the manuscript is different from the article information.

- The article title was between the first version and this current revised version. In the article information the original title is therefore different from the current version.

4. Page 4 line 100, the authors claimed that “General bibliometric research on the optimization of composite materials reveals a growing trend in research activity”. However, it is hard to tell from Figure 5 since there were less than 10 documents per year in the last 8 years. I believe that it is an upward trend, but your search keyword of “aircraft” might filter out many relevant articles.

- Thank you for your constructive comment. In fact, the number of articles and the sample size is too small to demonstrate this trend. The filter was expanded to include airspace structures, helicopters and drones.

5. The abbreviation of machine learning (ML) is defined twice (line 121 and line 165). Same issue for neural networks (NNs).

- Thank your sharp review. This error escaped our review. This correction was implemented in the text.

6. On page 7 line 162, the statement “DNNs are a specific type of NNs, however, DNNs have more hidden layers compared to NNs” is misleading. They are not comparable since DNN is a sublet of NN. In fact, it is debatable whether to have a separate section for DNN as some references before Section 3.1 use DNN with more than 3 hidden layers, for example, [20]. Most modern NN applications are deep based on this definition.

- Thank you for your constructive comment. These changes were implemented in the section 3 (letters in red).

7. On page 10 line 288, “Optimizing an aircraft is extremely important, as this is a fundamental practice for improving it”. Do you repeat the same statement twice?

- Thank you for your constructive comment. In fact, this statement confusing, since we are using synonyms optimizing-improving and important – fundamental. This sentence was removed.

8. Also on page 10, the subsection number should start with 5.1.

- Thank your sharp review. This typo escaped our review and was an error on the latex programming.  This correction was implemented in the text (letters in red).

9. The section title of section 5 (Survey of Previous Literature) can be confusing as the entire paper is about the previous work.

- Thank you for your constructive comment. We removed “of Previous” from the title (letters in red).

10. It is not necessary to have Sections 6 and 7. I would recommend moving those contents to Sections 5 and 8.

- Thank you for your constructive comment. We agree and changed accordingly. Section 6 is now section 5.7. The text of section 7 passed to section 8.

11. Further language polishing is recommended.

Round 2

Reviewer 1 Report (Previous Reviewer 2)

Comments and Suggestions for Authors

The authors corrected my comments and revised the article.

There are a couple more small comments:

1. Expand the abstract a little more and present more of the results obtained in it.

2. It would be better to rename the Discussion and Conclusions section to simply Conclusions or divide it into 2 sections.

Author Response

1. Expand the abstract a little more and present more of the results obtained in it.

 

Thank you for your constructive comment. The abstract has been expanded to present more about the results, as recommended.

 

2. It would be better to rename the Discussion and Conclusions section to simply Conclusions or divide it into 2 sections.

 

Thank you for your constructive comment. The section’s name has been changed to simply Conclusion, as recommended.

Reviewer 3 Report (New Reviewer)

Comments and Suggestions for Authors

The authors have addressed all my concerns and the manuscript is now ready for publication in present form.

Author Response

The authors have addressed all my concerns and the manuscript is now ready for publication in present form.

 

Thank you for your comment's

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This is an interesting paper and reviews the application of machine learning (ML) and optimization methods in aeronautical engineering, focusing on aircraft structures and composite materials. It highlights the growing application of ML methods, such as neural networks (NNs) and deep neural networks (DNNs), in improving aircraft performance, reducing costs, and enhancing material properties. Additionally, the paper discusses various optimization techniques, like genetic algorithms (GAs), that are integrated with ML to find the best solutions for complex engineering problems. The analysis shows research trends and the growing impact of these technologies in recent years. Overall, it is a good review paper. However, the paper needs some adjustments and some additional inputs to be ready for publication.

 

Point 1.

Ensure consistent use of terminology throughout the paper. For instance, once you introduce an abbreviation like ML (Machine Learning), use it consistently. Also, for "Neural Networks (NN)," and "Deep Neural Networks (DNN)" after their first introduction ensure consistent use of them. Make sure to double-check it for other cases as well.

Point 2.

Discussion on optimization techniques like Genetic Algorithms (GA) and their integration with ML is insightful. More examples and case studies could strengthen the argument (consider adding specific case studies or practical examples where ML has significantly impacted aircraft structure optimization).

Point 3.

It would be interesting to add a brief section on future research directions or emerging trends that would provide a forward-looking perspective.

 

 

 

Comments on the Quality of English Language

Minor editing of English language required

Author Response

Reviewer comment 1 – “This is an interesting paper and reviews the application of machine learning (ML) and optimization methods in aeronautical engineering, focusing on aircraft structures and composite materials. It highlights the growing application of ML methods, such as neural networks (NNs) and deep neural networks (DNNs), in improving aircraft performance, reducing costs, and enhancing material properties. Additionally, the paper discusses various optimization techniques, like genetic algorithms (GAs), that are integrated with ML to find the best solutions for complex engineering problems. The analysis shows research trends and the growing impact of these technologies in recent years. Overall, it is a good review paper. However, the paper needs some adjustments and some additional inputs to be ready for publication.

Point 1.

Ensure consistent use of terminology throughout the paper. For instance, once you introduce an abbreviation like ML (Machine Learning), use it consistently. Also, for "Neural Networks (NN)," and "Deep Neural Networks (DNN)" after their first introduction ensure consistent use of them. Make sure to double-check it for other cases as well.”

Response 1 – Thank you for pointing this out. We agree with this comment and therefore we have reviewed the abbreviations and its consistent use throughout the text. Changes in the text are marked (in red and yellow)

Reviewer comment 2 – “Point 2.

Discussion on optimization techniques like Genetic Algorithms (GA) and their integration with ML is insightful. More examples and case studies could strengthen the argument (consider adding specific case studies or practical examples where ML has significantly impacted aircraft structure optimization).”

Response 2 – Thank you for pointing this out. We agree with this comment and therefore we included, in the section concerning structural optimization, some paragraphs describing other works and references. We believe this will strengthen the argument.

Reviewer comment 3 – “Point 3.

It would be interesting to add a brief section on future research directions or emerging trends that would provide a forward-looking perspective.”

Response 3 – Thank you for pointing this out. We agree with this comment and therefore we added a brief section concerning future research directions and in conclusions we added one paragraph about this topic.

Reviewer 2 Report

Comments and Suggestions for Authors

The article is devoted to methods of machine learning and optimization of aircraft structures and composite materials: a brief overview

However, there are some comments regarding the work:

1. The Abstract section needs to be rewritten and shortened, reflecting the relevance of the problem being solved and the scientific novelty of the result obtained.

2. In the Abstract section, abbreviations should be excluded, for example (CFD, FE, etc.), referring to it in the text of the manuscript.

3. The introduction section should indicate the novelty of the research being conducted. At the end of the introduction section, it is necessary to define the purpose of the scientific research, and provide a detailed structure of the article with a presentation of the problems to be solved in the following sections.

4. The manuscript should consider machine learning broader than its use in neural networks. Since machine learning is a class of artificial intelligence methods, the characteristic feature of which is not the direct solution of a problem, but learning through the application of solutions to many similar problems. To construct such methods, the tools of mathematical statistics, numerical methods, mathematical analysis, optimization methods, probability theory, graph theory, and various techniques for working with data in digital form are used.

5. Figures 9 – Figures 12 are Christomatic in nature and do not carry a scientific load. They should be replaced with structures that illustrate specific machine learning models.

6. It would be appropriate to highlight the correlation of using machine learning for

7. predicting loads from data volumes related to various scenarios and in relation to the type of aircraft being studied.

8. Which genetic algorithms are most effective in determining the optimization of aircraft structures and composite materials?

9. Conclusions must be structured, highlighting the main scientific and especially practical results obtained, as well as recommendations for designers and mechanical engineers.

Author Response

Reviewer comment 1 – “The article is devoted to methods of machine learning and optimization of aircraft structures and composite materials: a brief overview

However, there are some comments regarding the work:

1. The Abstract section needs to be rewritten and shortened, reflecting the relevance of the problem being solved and the scientific novelty of the result obtained.”

Response 1 – Thank you for pointing this out. We agree with this comment and therefore the abstract was rewritten and shortened as suggested. Changes in the text are marked (in red and yellow)

Reviewer comment 2 – “2. In the Abstract section, abbreviations should be excluded, for example (CFD, FE, etc.), referring to it in the text of the manuscript.”

Response 2 – Thank you for pointing this out. We agree with this comment and therefore all the abbreviations were excluded from the abstract.

Reviewer comment 3 – “3. The introduction section should indicate the novelty of the research being conducted. At the end of the introduction section, it is necessary to define the purpose of the scientific research, and provide a detailed structure of the article with a presentation of the problems to be solved in the following sections.”

Response 3 – Thank you for pointing this out. We agree with this comment and therefore the introduction section has been restructured and changes have been made to respond the reviewer suggestion.

Reviewer comment 4 – “4. The manuscript should consider machine learning broader than its use in neural networks. Since machine learning is a class of artificial intelligence methods, the characteristic feature of which is not the direct solution of a problem, but learning through the application of solutions to many similar problems. To construct such methods, the tools of mathematical statistics, numerical methods, mathematical analysis, optimization methods, probability theory, graph theory, and various techniques for working with data in digital form are used.”

Response 4 – Thank you for pointing this out. We agree with this comment and therefore the title was changed to be more in line with the more specific use of neural networks. Changes were also made to the text to make it more specific and highlight the use of NNs.

Reviewer comment 5 – “5. Figures 9 – Figures 12 are Christomatic in nature and do not carry a scientific load. They should be replaced with structures that illustrate specific machine learning models.”

Response 5 – Thank you for pointing this out. But these figures are used to simply illustrate the description on the text. However, these are inspired on figures published in previous works.

Reviewer comment 6 – “6. It would be appropriate to highlight the correlation of using machine learning for

7. predicting loads from data volumes related to various scenarios and in relation to the type of aircraft being studied.”

Response 6 – Thank you for pointing this out. We agree with this comment and therefore we added a paragraph, in the structure optimization section to highlight the correlation.

Reviewer comment 7 – “8. Which genetic algorithms are most effective in determining the optimization of aircraft structures and composite materials?”

Response 7 – Thank you for pointing this out. However, the aim of this revision was to describe some examples, applied to aeronautical design and structural optimization, regarding the use of Genetic Algorithms with Neural networks. We are unable to determine the most effective on based on comparisons of different works.

Reviewer comment 8 – “9. Conclusions must be structured, highlighting the main scientific and especially practical results obtained, as well as recommendations for designers and mechanical engineers.”

Response 8 – Thank you for pointing this out. The conclusion section was restructured as requested. Now the conclusion shows recommendations for designer and mechanical engineers, as well highlight scientific and practical results obtained.

Reviewer 3 Report

Comments and Suggestions for Authors

With all respect due to authors, what they did is not at all enough.

This work is a kind of mess, considering that even from general point of views it does not fit criteria required for a short review, and nor does it particularly cover the filed of research. I do recommend against publication of this work.

Starting from title, what it claims and what text shows are pinpointing a disparity, to me as a reviewer it appears that they are written by different teams from different fields!

The introduction is a mess, fantasy, with only five citations to Scopus. There is no literature review on the importance and significance of AI in aircraft
structures and composite material. Aircraft structures and composite materials are not plausible indeed! A part of aircraft is composite materials, but searching with these two keywords does not lead to a real image of the filed. This idea looks abstract, and cannot be accepted as a title and literature survey. So, it is not introduction, but is (I mean it could be) analysis of literature. considering wrong choice of composite materials and aircraft structure, it is not worthy of consideration.

Then, ML is explained at a very low-level, which is meaningful to the reviewer. It is just copy and paste, considering bulky references.

I cannot see anything about aircraft structure, which looks weird to me.

This draft is not even an initial draft. Excluding (wrong) publication analysis, only 10 papers are taken from aircraft-related works among 55, such that mostly general papers are cited.

Is it written by AI or human being?!

Author Response

Reviewer comment 1 – “With all respect due to authors, what they did is not at all enough.

This work is a kind of mess, considering that even from general point of views it does not fit criteria required for a short review, and nor does it particularly cover the filed of research. I do recommend against publication of this work.

Starting from title, what it claims and what text shows are pinpointing a disparity, to me as a reviewer it appears that they are written by different teams from different fields!

The introduction is a mess, fantasy, with only five citations to Scopus. There is no literature review on the importance and significance of AI in aircraft
structures and composite material. Aircraft structures and composite materials are not plausible indeed! A part of aircraft is composite materials, but searching with these two keywords does not lead to a real image of the filed. This idea looks abstract, and cannot be accepted as a title and literature survey. So, it is not introduction, but is (I mean it could be) analysis of literature. considering wrong choice of composite materials and aircraft structure, it is not worthy of consideration.

Then, ML is explained at a very low-level, which is meaningful to the reviewer. It is just copy and paste, considering bulky references.

I cannot see anything about aircraft structure, which looks weird to me.

This draft is not even an initial draft. Excluding (wrong) publication analysis, only 10 papers are taken from aircraft-related works among 55, such that mostly general papers are cited.

Is it written by AI or human being?!”

Response 1 – Thank you for pointing out these limitations. To address these issues, the title was modified to better suit the main aim of this review, which is to show the use of neural networks and optimization techniques to optimize aircraft structures and composite materials. The introduction section was restructured and more citations were added. A small section on future research directions was added and a paragraph regarding this topic was added in the conclusions. Regarding your comment of the few articles on aircraft structure optimization, it was difficult to find articles on this topic; articles were found on structures from different areas and on the use of composite materials in different areas, as can be seen in Figures 9 and 11. This highlights the importance of more studies in the aeronautical area using the tools presented.