An Aircraft Trajectory Prediction Method Based on Trajectory Clustering and a Spatiotemporal Feature Network

Round 1

Reviewer 1 Report

In general, although nolvety in algorithms is limited, the applications are quite interesting.

I don't understand the necessity of using CNNs. And from your results, I cannot see any significant improvements when using CNNs. Have you ever try BiLSTM+attention? Please clarify the necessity of using CNNS.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper presents a method based on trajectory clustering and a spatio-temporal feature network for the prediction of aircraft trajectories. The method presented by the authors involves assessing automatic dependent surveillance broadcast data and uses Hausdorff distance and clustering with K-Medoids. A trajectory spatiotemporal feature extraction network and BiLSTM network is then constructed from the clustering results. The authors show improved trajectory prediction accuracy with their method in comparison to other existing algorithms. The article is generally well structured and the topic of increasing aircraft trajectory prediction accuracy is relevant for air traffic control efficiency and ensuring safe navigation. 

I have some suggestions for the authors to address:

Line 248: Rather than discussing the purpose of the paper here, since this was already described in the introduction section, I would recommend the authors instead discuss the purpose of section 3.3.

 

Line 12, line 15: the last segment of these sentences are fragments. Consider revising the sentence

 

Line 29: Grammar error, consider revising to: "International Civil Aviation Organization (ICAO) uses flight trajectorIES as the..."

 

Line 35: Grammar error, consider revising to: "Therefore, it is crucial to study accurate aircraft trajectory prediction."

 

Line 47-49: Grammar error, consider revising to: "Qian [10] et al. proposed an aircraft trajectory prediction model based on A back propagation neural net-48 work (BP)."

 

Line 56-58, 163-166, 230-233, 415-419, 454-459: this sentence is very long and is a bit difficult to follow. Consider dividing the sentence. 

 

Line 142, 242-243, 352-353, 407-409, 430-432, 450-453: These sentences are fragments. Consider revising.

 

Line 143: Grammar check, consider revising to: "Secondly, noisy trajectories are removed."

 

Figure 6 and 7: I recommend using transparent lines to show the cluster weights more clearly. 

 

Line 281: Grammar check, consider revising to: "THE LSTM model solves THE gradient loss and gradient explosion problem in the RNN model."

 

Line 307: Grammar check, consider revising to: "THE Attention mechanism [27] has been widely used in neural networks recently."

 

Line 313: Grammar check, consider revising to: "The calculation formulas are as followS"

 

Line 325: This line is unfinished. Please revise. 

 

Line 346-348: Grammar error on this sentence. Please revise or describe more thoroughly 

 

Figure 13: Should be mentioned in the text prior to showing the figure. 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf