Uncertainty quantification and reduction in aircraft trajectory prediction using Bayesian-Entropy information fusion. (August 2021)

• Record Type: Journal Article
• Authors: Wang, Yuhao; Pang, Yutian; Chen, Oliver; Iyer, Hari N.; Dutta, Parikshit; Menon, P.K.; Liu, Yongming
• Is Part Of: Reliability engineering & system safety. Volume 212(2021)
• Publication Date: 2021-08
• Subjects: Air traffic management -- Bayesian-Entropy method -- trajectory prediction -- uncertainty; Reliability (Engineering) -- Periodicals; System safety -- Periodicals; Industrial safety -- Periodicals; Fiabilité -- Périodiques; Sécurité des systèmes -- Périodiques; Sécurité du travail -- Périodiques; 620.00452
• DOI: 10.1016/j.ress.2021.107650
• Languages: English
• ISSNs: 0951-8320

Abstract

Highlights: Quantifies the uncertainty of GNATS simulation in en-route trajectory as randomness in waypoints coordinates. Information fusion of observation data and physical constraints to predict aircraft trajectory Encode physical constraints in trajectory simulation as a Bayesian-Entropy model selection framework Predicts runway overrun based on historical data from Sherlock Data Warehouse Abstract: Eliminating accidents while maintaining the integrity of the National Airspace System is one of the central objectives of the Next Generation Air Transportation System. This paper presents a Bayesian framework for accurate trajectory and accident prediction in National Airspace System using a high-fidelity trajectory simulation platform. Various uncertainties in aircraft trajectory prediction due to pilot behavior and weather effects are included as random variables in the simulations. Bayesian-Entropy method fuses available observation data (e.g., positioning system) with existing physical constraints (e.g. runway location) to update these simulation parameters. The posterior distributions of parameters are used to predict the probability of an adverse incident and time-remaining to incident. The proposed Bayesian updating scheme offers a flexible and rigorous way for adverse incident diagnostics and prognostics in current and future Air Traffic Management. Two realistic examples are given to show that it is possible to derive advance warning using the proposed methodology. The approach integrates data from a simulation model, with real-time traffic data streams and available physical constraints, using the Bayesian-Entropy information fusion methodology. This advance warning will allow the pilots/controllers to take actions to mitigate adverse incident in the National Airspace System.