Abstract

With the advent of the big data era of air traffic control systems, the application of trajectory clustering in the field of air traffic has received widespread attention. This article reviews the trajectory clustering methods proposed in domestic and foreign literature. According to the different similarity measures and clustering evaluation criteria, the application of different clustering methods in trajectory clustering is introduced from several aspects, and the advantages and applicable scenarios of each algorithm are summarized. Then, based on the real flight history radar data information, the performance of the two different clustering algorithms in track clustering is analyzed. It mainly includes two aspects: First, in the data preprocessing part, data cleaning, filtering, and interpolation are performed. After processing, the data is resampled to obtain research data. Then, according to the characteristics of the track data, the two hyperparameters of the DBSCAN algorithm and the K value of the K-means algorithm are determined, and the clustering results are visually displayed in combination with the real flight data from the Capital International Airport to Shanghai Hongqiao International Airport.

Highlights

  • With the rapid economic development, the civil aviation industry is developing rapidly

  • According to different similarity measures and clustering evaluation criteria, the existing spatial clustering algorithms can be roughly divided into: clustering methods based on division, clustering methods based on hierarchy, clustering methods based on density, clustering based on grids Class methods, model-based clustering methods and fuzzy clustering algorithms, etc

  • In order to make the trajectory set more suitable for clustering, all the trajectories after interpolation processing are resampled into equal-length sequences, so that each trajectory contains the same number of radar points. the data preprocessing process is shown in Fig 1. the data after data preprocessing can be used for the study of track clustering

Read more

Summary

Introduction

With the rapid economic development, the civil aviation industry is developing rapidly. Comprehensive control operation data such as trajectory data, flight plan data, voice data, etc., through the mining and analysis of massive operation data, it is possible to realize uninterrupted real-time performance of safety-related indicators such as air traffic flow situation, control operation safety, and workload Monitoring and evaluation This will help industry personnel understand the characteristics of traffic flow in the airspace, assess the traffic potential of the route, and is of great significance to alleviating the pressure of air traffic congestion and ensuring the safe and efficient operation of aircraft. This article reviews and summarizes the trajectory clustering methods proposed in domestic and foreign literature Classify and compare these algorithms, find the differences between them, analyze the differences, and put forward some research conclusions. By using two different clustering methods to analyze the real radar track data, the difference between the two is intuitively reflected

Overview of track clustering algorithm
Clustering based on partition
Hierarchical clustering
Density-based clustering
Grid-based clustering
Model-based clustering
Fuzzy clustering algorithm
Data introduction
Data cleaning
Interpolation of missing data
K-means algorithm
DBSCAN algorithm
Data set
Parameter selection
Cluster performance analysis
Findings
Conclusions
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call