Ship Traffic Flow Analysis and Prediction in High-Traffic Areas Under Complex Environments
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe manuscript “Ship Traffic Flow Analysis and Prediction in High-Traffic Areas under Complex Environments” presents a hybrid prediction model (WVMA-LSTM) that combines Variational Mode Decomposition (VMD), Whale Optimization Algorithm (WOA).
The main goal of the study, as interpreted from the text, is to improve the accuracy and reliability of ship traffic forecasting by integrating signal decomposition and optimization with deep learning methods. However, the objective is not stated clearly in one unified sentence — it is spread across several paragraphs in the introduction and methodology. The paper would benefit from an explicit statement of purpose.
The overall structure of the manuscript is logical and consistent with the MDPI format. Nonetheless, it would help readers if the authors added a short paragraph at the end of the Introduction outlining the paper’s organization (e.g., “The remainder of this paper is organized as follows…”). While this is not a scientific issue, it significantly improves readability.
In terms of content, several areas require attention before the paper can be accepted:
1. The authors mention that AIS data come from three U.S. regions, but no details are given about data sources, access, or cleaning methods. Applied Sciences requires a higher level of reproducibility; the statement “dataset available upon request” is insufficient. Please specify the data origin, time range, and preprocessing steps.
2. The WVMA-LSTM method is sound, but Sections 3.1–3.4 include long theoretical descriptions (nearly ten pages) that could be shortened. Focus on implementation logic rather than re-deriving known algorithms.
3. The analysis focuses mainly on R² and RMSE values, but lacks discussion of their practical implications. For instance, how would improved prediction accuracy affect lock scheduling, congestion management, or navigational safety?
4. Figures are cluttered and contain embedded labels and unnecessary graphical elements. MDPI standards favor simple, clean visuals with readable captions. Figures should be redrawn for clarity. Some of the information currently embedded in the figures—particularly in Figure 3—could be moved to the captions, as the present layout is cluttered and difficult to read.
5. Section 5 (“Conclusions”) is technically correct but lacks a Limitations paragraph. It would be helpful to acknowledge that the model is computationally intensive and sensitive to the quality of meteorological inputs.
Author Response
Thank you very much for your valuable suggestion. According to your comment, we have added a short paragraph at the end of the Introduction to outline the structure of the paper. Other related revisions can be found in the attached revised manuscript. Please see the attachment for details.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThe authors have developed a novel forecasting approach based on WVMA-LSTM, which involves decomposing vessel traffic time series before forecasting. The proposed model consists of three main components: 1. Vessel traffic data are first decomposed using the variational mode decomposition (VMD) method. 2. The Pearson correlation coefficient (PCC) is used 3. A long short-term memory (LSTM) model combined with a multi-head attention mechanism. The article was well-written and was read with interest. The strength of this study lies in the developed method, which is based on a combined use of a hybrid prediction model (WVMA-LSTM) that incorporates the Whale Optimization Algorithm (WOA), Variational Mode Decomposition (VMD), and a multi-head attention mechanism.
However, the article could benefit from some improvements.
1. The literature review could be enhanced. Given that this study aims to improve traffic flow in high-traffic areas, using evolutionary techniques such as the Whale Optimization Algorithm would be beneficial. It would also be useful to mention and cite some recent works on evolutionary competitions:
[1] A. S. Akopov and L. A. Beklaryan, "Evolutionary Synthesis of High-Capacity Reconfigurable Multilayer Road Networks Using a Multiagent Hybrid Clustering-Assisted Genetic Algorithm," in IEEE Access, vol. 13, pp. 53448-53474, 2025, doi: 10.1109/ACCESS.2025.3554054.
[2] Y. -T. Tseng and H. -W. Ferng, "An Improved Traffic Rerouting Strategy Using Real-Time Traffic Information and Decisive Weights," in IEEE Transactions on Vehicular Technology, vol. 70, no. 10, pp. 9741-9751, Oct. 2021, doi: 10.1109/TVT.2021.3102706. .
2. Section 3. "System Architecture and Key Algorithms" could be improved. On page 5, the pseudocode "Input: Original Time Series" is presented. This pseudocode, in particular, uses variables such as x0(t), x1(t), etc. However, these notations should be introduced prior to the first use in the text, that is, before the pseudocode.
3. When formulating the optimization problem (Eq. (1)), the authors introduce the Lagrange multiplier α(t). However, it is not clear how this variable is used later on. It is also not completely clear how the u(t) variable with a top cover and iterator n differs from the u(t) variable introduced earlier as the Kth Intrinsic Mode Function (IMF) in Eq. (2). The authors are advised to carefully review all formulas and notations used in the paper's text.
4. The article has several issues with the incorrect formatting and alignment of mathematical equations, figures, and tables (e.g., see P. 6,15).
5. The quality of Figures 9 and 10 is substandard, particularly due to the extremely small font used for the legends' text.
6. The findings of the study indicate that the WOA-VMD-LSTM outperforms the WOA-EMD-LSTM in all three datasets, demonstrating the superiority of the former. However, it should be noted that there are various evolutionary algorithms that can be combined with the LSTM model, such as genetic algorithms and particle swarm optimization. It is not yet clear whether WOA-VMD-LSTM would outperform GA-VMD-LSTM or PSO-VMD-LSTM. The paper should provide a more in-depth discussion of potentially alternative implementations of the framework proposed by the authors.
Author Response
Thank you very much for your valuable suggestion.Please see the attachment part
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsThere's a noticeable improvement in quality, making paper acceptable. Editing notes will be added in the next steps.
Reviewer 2 Report
Comments and Suggestions for AuthorsThe article has been significantly improved by the authors based on the recommendations of reviewers. This study contributes to the development of a new prediction framework that decomposes vessel traffic flow data prior to forecasting. The framework enables the analysis and prediction of ship traffic flow in high-traffic areas under complex environmental conditions. The paper is recommended for publication.
