Combining Machine Learning Models to Improve Estimated Time of Arrival Predictions
All aviation stakeholders require accurate estimated times of arrival in order to run flight operations as efficiently as possible. The time of arrival, however, is difficult to predict because it is affected by the uncertainties of the previous flight phases, with take-off time variability being t...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
TU Delft OPEN Publishing
2025-01-01
|
| Series: | European Journal of Transport and Infrastructure Research |
| Subjects: | |
| Online Access: | https://journals.open.tudelft.nl/ejtir/article/view/7488 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1841545229466861568 |
|---|---|
| author | Ramon Dalmau Aymeric Trzmiel Stephen Kirby |
| author_facet | Ramon Dalmau Aymeric Trzmiel Stephen Kirby |
| author_sort | Ramon Dalmau |
| collection | DOAJ |
| description |
All aviation stakeholders require accurate estimated times of arrival in order to run flight operations as efficiently as possible. The time of arrival, however, is difficult to predict because it is affected by the uncertainties of the previous flight phases, with take-off time variability being the most significant contributor. At present, estimated time of arrival predictions are computed by the Enhanced Traffic Flow Management System, which collects data from a variety of sources to provide the best estimate throughout the entire duration of the flight. This paper introduces a novel approach that leverages existing machine learning models to enhance the accuracy of estimated time of arrival predictions, also during the pre-departure phase. More specifically, the first model (Knock-on) anticipates rotational reactionary delays arising from unrealistic available turn-around times; the second model (FADE) forecasts the evolution of air traffic flow management delays for regulated flights; and the third model, AirborneTime, was trained to identify systematic discrepancies between reported and actual airborne times. Using a dataset comprised of historical traffic and meteorological data collected during one year, this paper presents a comprehensive evaluation of this ensemble of models, referred to as PETA, against the current predictions across various time horizons, ranging from 6 hours before departure to the moment of take-off. The results indicate that the proposed solution surpasses the existing system in approximately two-thirds of the predictions. When the proposed solution performs better, the average and median improvements are 14 minutes and 7 minutes, respectively. However, when it underperforms, the average and median deteriorations are 7 minutes and 4 minutes, respectively. The optimal time frame appears to be between 2 and 6 hours before the departure time. This quantitative data is supported by feedback from European airlines, air navigation service providers and airports who used PETA in a live trial.
|
| format | Article |
| id | doaj-art-c53f31e65b1c4da8a8df1bca78da03e5 |
| institution | Kabale University |
| issn | 1567-7141 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | TU Delft OPEN Publishing |
| record_format | Article |
| series | European Journal of Transport and Infrastructure Research |
| spelling | doaj-art-c53f31e65b1c4da8a8df1bca78da03e52025-01-12T09:43:45ZengTU Delft OPEN PublishingEuropean Journal of Transport and Infrastructure Research1567-71412025-01-0125110.59490/ejtir.2025.25.1.7488Combining Machine Learning Models to Improve Estimated Time of Arrival PredictionsRamon Dalmau0Aymeric Trzmiel1Stephen Kirby2EUROCONTROLEUROCONTROLEUROCONTROL All aviation stakeholders require accurate estimated times of arrival in order to run flight operations as efficiently as possible. The time of arrival, however, is difficult to predict because it is affected by the uncertainties of the previous flight phases, with take-off time variability being the most significant contributor. At present, estimated time of arrival predictions are computed by the Enhanced Traffic Flow Management System, which collects data from a variety of sources to provide the best estimate throughout the entire duration of the flight. This paper introduces a novel approach that leverages existing machine learning models to enhance the accuracy of estimated time of arrival predictions, also during the pre-departure phase. More specifically, the first model (Knock-on) anticipates rotational reactionary delays arising from unrealistic available turn-around times; the second model (FADE) forecasts the evolution of air traffic flow management delays for regulated flights; and the third model, AirborneTime, was trained to identify systematic discrepancies between reported and actual airborne times. Using a dataset comprised of historical traffic and meteorological data collected during one year, this paper presents a comprehensive evaluation of this ensemble of models, referred to as PETA, against the current predictions across various time horizons, ranging from 6 hours before departure to the moment of take-off. The results indicate that the proposed solution surpasses the existing system in approximately two-thirds of the predictions. When the proposed solution performs better, the average and median improvements are 14 minutes and 7 minutes, respectively. However, when it underperforms, the average and median deteriorations are 7 minutes and 4 minutes, respectively. The optimal time frame appears to be between 2 and 6 hours before the departure time. This quantitative data is supported by feedback from European airlines, air navigation service providers and airports who used PETA in a live trial. https://journals.open.tudelft.nl/ejtir/article/view/7488machine learningflight predictabilityestimated time of arrival |
| spellingShingle | Ramon Dalmau Aymeric Trzmiel Stephen Kirby Combining Machine Learning Models to Improve Estimated Time of Arrival Predictions European Journal of Transport and Infrastructure Research machine learning flight predictability estimated time of arrival |
| title | Combining Machine Learning Models to Improve Estimated Time of Arrival Predictions |
| title_full | Combining Machine Learning Models to Improve Estimated Time of Arrival Predictions |
| title_fullStr | Combining Machine Learning Models to Improve Estimated Time of Arrival Predictions |
| title_full_unstemmed | Combining Machine Learning Models to Improve Estimated Time of Arrival Predictions |
| title_short | Combining Machine Learning Models to Improve Estimated Time of Arrival Predictions |
| title_sort | combining machine learning models to improve estimated time of arrival predictions |
| topic | machine learning flight predictability estimated time of arrival |
| url | https://journals.open.tudelft.nl/ejtir/article/view/7488 |
| work_keys_str_mv | AT ramondalmau combiningmachinelearningmodelstoimproveestimatedtimeofarrivalpredictions AT aymerictrzmiel combiningmachinelearningmodelstoimproveestimatedtimeofarrivalpredictions AT stephenkirby combiningmachinelearningmodelstoimproveestimatedtimeofarrivalpredictions |