A New Mid-range Conflict Detection Model And Associated Algorithm Based On The Unscented Particle Filter

Effective conflict detection and conflict resolution are crucial to the prevention ofpotential conflict, especially when the air traffic is becoming heavier. In the conflictdetection stage, the future positions of aircraft in an observed region can be predictedbased on their current positions and flight plans. If the predicted positions between anypair of aircraft violate safe separation requirements in the future, a potential conflictis declared, and the conflict resolution should be followed up. In the conflict reso-lution stage, the flight plans of aircraft involved in the conflict should be modified.These two stages should be performed until there is no conflict in the observed region.Conflict detection and resolution are performed at three different levels within long,mid, and short ranges of the air traffic management process. The work presented hereconcentrates primarily on the mid-range conflict detection.In practice, aircraft motion is affected by many factors, which will result in errorbetween aircraft trajectories and its planed airway. Hence, the predictions of aircrafttrajectories are inexact. A probabilistic model for the global effect of the perturbationsentering the aircraft motion was proposed by Paielli and Erzberger in 1997. In 1999and 2000, Prandini, Lygeros, Nilim et al. studied mid-range conflict detection in termsof this probabilistic model.To increase the accuracy of mid-range conflict detection, designing a more ac-curate probabilistic model of aircraft motion is necessary. At first, a nonlinear anddiscrete probabilistic model is developed to describe aircraft motion system. Becauseobservation error of radar is taken into account, estimation of probability of conflictshifts to calculation of posterior probability density of state. Theoretically, the particlefilter can solve this question better. The study used unscented particle filter to estimate...