Probabilistic Prediction of Separation Buffer to Compensate for the Closing Effect on Final Approach

Abstract

The air traffic is mainly divided into en-route flight segments, arrival and departure segments inside the terminal maneuvering area, and ground operations at the airport. To support utilizing available capacity more efficiently, in our contribution we focus on the prediction of arrival procedures, in particular, the time-to-fly from the turn onto the final approach course to the threshold. The predictions are then used to determine advice for the controller regarding time-to-lose or time-to-gain for optimizing the separation within a sequence of aircraft. Most prediction methods developed so far provide only a point estimate for the time-to-fly. Complementary, we see the need to further account for the uncertain nature of aircraft movement based on a probabilistic prediction approach. This becomes very important in cases where the air traffic system is operated at its limits to prevent safety-critical incidents, e.g., separation infringements due to very tight separation. Our approach is based on the Quantile Regression Forest technique that can provide a measure of uncertainty of the prediction not only in form of a prediction interval but also by generating a probability distribution over the dependent variable. While the data preparation, model training, and tuning steps are identical to classic Random Forest methods, in the prediction phase, Quantile Regression Forests provide a quantile function to express the uncertainty of the prediction. After developing the model, we further investigate the interpretation of the results and provide a way for deriving advice to the controller from it. With this contribution, there is now a tool available that allows a more sophisticated prediction of time-to-fly, depending on the specific needs of the use case and which helps to separate arriving aircraft more efficiently.

Predictive classification and understanding of weather impact on airport performance through machine learning

“…The propagation of delay in the network is paramount when assessing the impact of (local) congestion (Campanelli et al, 2016;Ivanov et al, 2017;Baspinar et al, 2016). Particularly, research is conducted to evaluate the impact of disruptions on airport turnaround operations (Postorino et al, 2020), the impact of sudden and slow onset weather events on departure delays (Borsky and Unterberger, 2019), the handling of uncertainties in the arrival management Förster et al, 2021), and the management of airports in extreme winter conditions (Merkert and Mangia, 2012). The delay propagation is particularly critical when estimating the resilience of the air traffic management system and the impact of different mechanisms on the expected performances' variations (Cook et al, 2016;Proag and Proag, 2014;Cook et al, 2009).…”

Probabilistic Flight Delay Predictions Using Machine Learning and Applications to the Flight-to-Gate Assignment Problem

“…However, there are very few examples of studies utilizing the algorithm to obtain probability distributions. Förster et al [34] use quantile values, obtained from Quantile Random Forests, to construct a right-continuous cumulative distribution function of aircraft's time-to-fly from the turn onto the final approach course to the runway threshold. Schlosser et al [35] and Rahman et al [36] use Random Forests algorithms to obtain probability distributions for precipitation forecasts and drug sensitivity, respectively.…”

Implementation of a Long-Range Air Traffic Flow Management for the Asia-Pacific Region

“…The flexible use of airspace [38], [39] and high utilization of the runway system [40]- [42] are key elements to ensure efficient use of the (declared) airport capacity, even under different weather constraints [9], [43]. Aircraft arrival processes are mainly distinguished by two parts; flow-based control (management of a group of flights) and time-based tactical control (individual sequencing) [4], [44].…”