Probabilistic Approach to Trajectory Conformance Monitoring

Trajectory conformance monitoring checks whether an aircraft is likely to deviate significantly from its contract trajectory in the near future. This paper presents a systematic procedure for the design and evaluation of probabilistic trajectory conformancemonitoring algorithms. Basic components of a conformancemonitoring system are decomposed into trajectory prediction, calculation of deviation metrics, and decision logics for declaring nonconformances. Likely causes of nonconformances are categorized and stochastic kinematic trajectory predictions are used. The deviation metric is defined as the probability with which predicted trajectories exceed a containment region around the contract trajectory. An algorithm is presented that can estimate the nonconformance probability over time for uncertainties with arbitrary probabilistic distributions. When this nonconformance probability reaches a predefined threshold within the decision interval, a nonconformance is declared. In this paper, performances of conformancemonitoring algorithms are evaluated in terms of rates of false alarm,missed detection, correct detection of nonconformance, and correct confirmation of conformance. In addition, the advance time from detection to occurrence of a nonconformance is also used to measure the timeliness of an alert in correct detections. Monte Carlo simulations of aircraft level turns are used to systematically study the effects of various algorithm parameters on conformance monitoring system performances.