A Bayesian Approach for Conformance Monitoring

Abstract

For safe and efficient management of air traffic, the prediction of aircraft trajectories into the future is necessary. Conformance monitoring is one of important functions in trajectory prediction to ensure that aircraft adhere to the assigned flight plan. One of challenges for timely detection of a non-conforming aircraft is that its trajectories cannot be fully specified due to uncertainty on how the aircraft executes the currently applied flight plan. This uncertainty, together with noisy measurements, can create a series of data poor situations which often results in a high level of false detection or detection delay. To overcome this problem, our paper proposes a new method of conformance monitoring by using a Bayesian formulation. A prior available information can be utilized in the proposed method to reduce false detections without an arbitrary increase in detection delay. Furthermore, the proposed method can provide actual probability of conformance which can bring various capabilities to trajectory prediction. The proposed method is illustrated with examples.Nomenclature z(k) measurement of aircraft's state at time k Z(k) a set of all measurements up to time k T CT trajectory change time for a non-conforming aircraft Ω a set two conformance statuses; ω 0 -conformance and ω 1 -non-conformance H a set of TCT hypotheses

Analysis of UTM Tracking Performance for Conformance Monitoring via Hybrid SITL Monte Carlo Methods

“…In particular, transitions pose a challenge in CM due to turning dynamics and ambiguity in turn initiation time. As an alternative to the fault-detection formulation, Lee et al developed a Bayesian approach for CM [24], in part to address ambiguity in transition time. At every timestep, the algorithm estimates conformance probability across a range of trajectory change times (TCTs), and calculates their weighted sum.…”