Using Hidden Markov Models to Detect Mode Changes in Aircraft Flight Data for Conflict Resolution

Abstract

Verification and validation of conflict detection and resolution aids is critical in the air traffic management domain. With increasing capacity and safety constraints, the automation of critical tasks to reduce controller workload becomes essential. Classic aircraft conflict resolution strategies involve separating projected conflicting aircraft via altitude. However, the ability to sense, process and actuate upon aircraft data is a non-trivial task. Models of aircraft trajectories are non-linear and stochastic in nature; and their internal parameters are often poorly defined. Verifiable techniques for mode detection are critical in order to enable the accurate projection of aircraft conflicts, and for the enactment of altitude separation resolution strategies. In this paper, we use learning techniques from hidden Markov modeling in order to tune the discrete and continuous parameters of a stochastic hybrid model in order to perform mode detection upon actual flight track data. Comparisons of aircraft in ascending, descending and level flight are performed, and unknown flight track data is evaluated probabilistically against the tuned model.