Risk Prediction of Shipborne Aircraft Landing Based on Deep Learning

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

A brief summary 

The topic of the proposed article (manuscript) is ensuring a safe landing of a fighter aircraft on an aircraft carrier using elements of artificial intelligence, including neural network technology and deep machine learning. The authors consider the aim of the article as filling the gap that exists in this topic compared to aircraft landings in ground conditions. To do this, they use adequate scientific methodology and examine relevant technical issues. The described study can contribute to improving the safety and efficiency of naval aviation operations. In this regard, the article is of undoubted importance.

General concept comments

Along with the relevance of the study and the scientific interest that this article represents, some of its shortcomings should be noted.

I. There is no definition of the concept of risk in the context under consideration, in particular “the risk of shipborne aircraft landings”. The indicators by which this risk is measured are also not described. Some vague idea is given only in  Conclusions, in paragraph 1.

II. In this article, neural network technology is actually considered as an alternative to ensuring a safe landing by calculating the parameters of the aircraft flight trajectory based on the equations of motion, using the methods of controlled flight dynamics in the atmosphere. It seems necessary to explain the reasons for such a change in the approach to ensuring safety. The introduction of new technologies may bring new risks, and this issue is not studied or even posed in the article.

III. It is difficult to understand from the article how landing control is supposed to be implemented using the studied machine learning methods: as a device on board an aircraft, on an aircraft carrier, as a distributed system, or as part of a flight simulator?

Specific comments that relate to individual items of the article are provided in the corresponding section of the attached file.

After the authors have made corrections to the manuscript that correspond to the above comments, or have given reasoned objections to them, the article can be accepted for publication.

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This paper presents a study on risk prediction for shipborne fighters using deep learning methods. While the work addresses an important application domain, there are several significant methodological and presentation issues that require substantial revision.

Major Issues

Limited Methodological Scope: The paper exclusively focuses on deep learning approaches without considering other well-established machine learning methods. The absence of gradient boosting methods, random forests, or other conventional ML approaches significantly limits the comprehensiveness of the study and makes it difficult to assess the relative performance of the proposed methods.

Insufficient Comparative Analysis: The primary limitation of this work is the lack of cross-method comparison. The experiments only evaluate different feature combinations within each model rather than comparing the performance of different algorithms. This approach provides limited insights into which methods are most suitable for the specific problem of shipborne fighter risk prediction.

Missing Baseline Methods: The study would benefit significantly from including conventional machine learning baselines. A comprehensive comparison should include both traditional ML methods and deep learning approaches to demonstrate the added value of the proposed deep learning solutions.

Technical Comments

Terminology Consistency (Section 2.1.2): Please clarify whether "key valuables" should be "key variables" throughout the text.

Table 1 Accuracy: There appears to be a mismatch between acronyms and their full names. For example, "Center deviation (DCA)" - please verify that all abbreviations correctly correspond to their definitions.

Figure Quality (Figure 4): The figure contains Chinese characters, which may not be accessible to all readers. Please provide an English version or ensure all text elements are in English.

Minor Issues

Abstract Scope: The abstract suggests that only deep learning methods are considered. If this is intentional, please justify why other proven methods (e.g., gradient boosting) were excluded from consideration.

Literature Review: The literature review section is well-written and comprehensive.

Recommendations for Improvement

 

Expand Methodological Framework: Include conventional ML methods such as gradient boosting, random forests, and support vector machines to provide a more complete comparison.

Implement Cross-Method Comparison: Design experiments that directly compare different algorithms on the same dataset and feature sets.

Reference Exemplary Work: Consider following the way of presentation demonstrated in Lui et al. (2025)*, which effectively combines feature engineering with method comparison: "Enhancing aircraft arrival transit time prediction: A two-stage gradient boosting approach with weather and trajectory features." Journal of the Air Transport Research Society, 4, 100062.

Address Technical Issues: Correct the terminology inconsistencies, acronym mismatches, and language issues identified above.

Decision Recommendation: Major revisions

The paper addresses an important application area, but the current methodological approach is too narrow and the experimental design lacks the comprehensive comparisons necessary for a strong contribution. The authors should substantially expand their methodology to include conventional ML approaches and redesign their experiments to enable meaningful cross-method comparisons before resubmission.

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have addressed my concerns.

Author Response

Please refer to the attachment.

Author Response File: Author Response.doc