Hybrid Machine Learning and Estimation-Based Flight Trajectory Prediction for Air Traffic Management

Researchers at Purdue University have developed a framework for trajectory projections in terminal airspace by combining a physics-based estimation method as well as a machine learning method. A trajectory prediction model based on machine learning is trained from historical surveillance data to represent the collective behavior of a set of flight trajectories, from which the data-driven prediction can be obtained as the expected future behavior of an incoming flight. A physics-based estimation algorithm called Residual-Mean Interacting Multiple Models then incorporates the machine learning prediction as a pseudo-measurement to account for the current motion of the aircraft. The proposed framework produces a greatly improved prediction accuracy compared to the two existing machine learning-based algorithms.

Technology Validation: This technology has been validated using real air traffic surveillance data and running the technology's prediction model.

Advantages:

-More accurate than the current in use model

-Enhances the air traffic controller's situational awareness.

-Increases safety and efficiency of air traffic operations.

Applications:

-Integration into the Federal Aviation Administration's air traffic management software

TRL: 4

Intellectual Property:

Keywords: Flight trajectory prediction, terminal airspace, machine learning in air traffic, physics-based estimation, Residual-Mean Interacting Multiple Models, air traffic management software, FAA integration, air traffic controller situational awareness, air traffic safety and efficiency, hybrid trajectory prediction, Aircraft trajectory prediction, Gaussian mixture model, Hybrid estimation, Interacting multiple models