Citation

Lu F, Chen Z, Chen H. Transp. Res. C Emerg. Technol. 2021; 124: e102970.

Copyright

(Copyright © 2021, Elsevier Publishing)

DOI

10.1016/j.trc.2021.102970

PMID

unavailable

Abstract

In order to ensure the safe operation of parallel routes in the ocean area, it is necessary to assess the risk of lateral collision of routes. The data used in the assessment often come from the real flights, while the traditional land-based ADS-B equipment has limited coverage and cannot be used for research. The space-based ADS-B system can receive and send the flight position of the aircraft over the ocean area. In this paper, the ocean tracks in space-based ADS-B data are first extracted. According to the different route structures, the data of fixed route and flexible route are preprocessed by segmented Kalman filter and cubic polynomial fitting method respectively. Then the processed track position error is calculated, and Q-Q (Quantile-Quantile) median method is used to test the normal distribution of the error, and k-means clustering is used to obtain the clustering center and occurrence frequency of positioning error and flight error. Finally, a collision risk model based on Bayesian network is established, and the error distribution parameters and occurrence frequency are brought into the model to evaluate the collision risk of parallel routes in the ocean area. The results show that the lateral collision risk based on space-based ADS-B ocean flight data is lower than 0.5 × 10−8 times/hour specified by ICAO when the fixed route separation is 14n miles and the flexible route separation is 49n miles. And the actual lateral interval of the Pacific flexible route is 50n miles (Jiang, 2010), which meets the safety target level.

Language: en

Keywords

Bayesian network; Collision risk; Machine learning; Segmented Kalman filter; Space-based ADS-B; Track fitting