No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
A Data-driven approach for taxi-time prediction: a case study of Singapore Changi airport
Abstract
In daily operations at an airport, the ground movement of an aircraft is one of the most critical airside operations. The ground movement problem includes two sub-problems: routing and scheduling, which serve the purpose of guiding aircraft on the surface of an airport to meet the departure schedule while minimizing overall travel time. Ground-movement controllers manage the taxi-route assignments and taxi-time estimation for each aircraft in the arrival or departure queue. A high-accuracy taxi-time calculation is required to increase the efficiency of airport operations. In this study, we propose a data-driven approach to construct features set and build predictive models for taxi-time prediction for departure flights. The proposed approach can suggest, both, taxi-route and predict the corresponding taxi-time: by analyzing ground movement data. The controller's operational preferences are extracted and learned by machine learning algorithms for predicting taxi-route and taxi-time of given aircraft. In this approach, we take advantage of taxiing trajectories to learn the controller's decision, which reflects how the controller had decided the routing for a given situation. Two machine learning models, random forest regression and linear regression are implemented and show similar performances in estimating the taxi-time, however, from our observations, the random forest model can provide a more stable result and interpretability which is suitable for real operations. The predictive model for taxi-time can predict the taxi-out time with high accuracy with a given assigned taxi-route. The model can cover the controller's decision up to 70% in the top-1 and 89% in top-2 recommends. The Mean Absolute Error is less than 2.07 minutes for all departure flights and Root Mean Square Error is approximately 2.5 minutes. Moreover, the ±3-minute error window can cover around 76% of departures while more than 95% of departures are within the ±5-minute error window.
... Zhi et al. [15] proposed a taxi-out time prediction model for departure flights based on local weighted support vector regression, and the prediction results show that the accuracy within ±3 min can reach 83.3%, which is better than the traditional support vector regression model. Pham et al. [16] performed the prediction in two stages, first predicting the taxi path of the flight and then predicting the taxi time according to the taxi path and taxi distance. The prediction results show that the accuracy within ±5 min can reach 95%. ...
... (1) The existing studies consider taxi time as a whole [4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19], while not analyzing UTT and ATT separately. (2) The existing studies consider the flow of the scene as a whole [4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19], without focusing on the flow associated with the microstructure of the scene, such as the flow of the corridor. ...
... (1) The existing studies consider taxi time as a whole [4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19], while not analyzing UTT and ATT separately. (2) The existing studies consider the flow of the scene as a whole [4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19], without focusing on the flow associated with the microstructure of the scene, such as the flow of the corridor. ...
The taxi-out time of an airport scene can be categorized into the unimpeded taxi-out time and the additional taxi-out time. Usually, additional taxi-out time is used as a key index to monitor taxi-out performance, and its accurate prediction plays an important role in optimizing the allocation of time slots at an airport and improving scene operation efficiency. Taking Shanghai Pudong International Airport as the research object, we first analyze its layout and construct the origin–destination pairs (ODPs) based on the stand groups and runways. Then, we develop a multiple linear regression model based on the arrival and departure flows to calculate the unimpeded taxi-out times for all ODPs. The actual taxi-out time is then subtracted from the unimpeded taxi-out time to obtain the historical additional taxi-out time of each flight. We propose three new flow features related to the structure: the corridor departure flow, the corridor arrival flow, and the departure flow proportion of ODPs, based on which we construct a dataset for training the prediction model. We then propose an additional taxi-out time prediction model based on the nutcracker optimization algorithm (NOA) and XGBoost and run comparison experiments on the operation data of our target airport. The results show that the optimized prediction model we proposed has the best performance compared with the traditional XGBoost model and other commonly used prediction models, and the proposed structure-related features have high correlations with additional taxi-out time.
... This is achieved by investing the impact of different factors, drawing conclusions on the most important factor for accurately modeling taxi time [5][6][7][8][9][10]. Another kind of data-driven approach is traditional machine learning, which is achieved by constructing feature sets that may affect aircraft taxi time and utilizing extensive historical operational data to establish taxi time prediction models [11][12][13][14][15][16][17][18]. However, these methods have some drawbacks: ...
... As traditional machine learning techniques continue to advance, particularly with regard to their ability to mine nonlinear relationships in data, they are seeing increasing application in studies [11][12][13][14][15][16][17][18] focused on predicting the taxi time of aircraft. In general, expert-defined feature sets [16,18] that may affect an aircraft's taxi time can be categorized into flight properties, airport operational information, traffic conditions, and weather conditions (the details of the influencing factors can be found in Appendix A). ...
... • RF. Random forests (RF), which have achieved a superior prediction performance in previous works [17,18], are used for aircraft's taxi time prediction. ...
Variable taxi time prediction is the core of the Airport Collaborative Decision Making (A-CDM) system. An accurate taxi time prediction contributes to enhancing airport operational efficiency, safety and predictability. The deep dynamic spatio-temporal correlation inherent in airport traffic data is critical for taxi time prediction. However, existing machine learning (deep learning) methods have been unable to thoroughly exploit these correlations. To address this issue, we propose a deep learning-based model called the multi-task dynamic spatio-temporal graph attention network (MT-DSTGAN). Our model also predicts future entire airport traffic flow and taxiing segment traffic flow as auxiliary tasks, with the goal of enhancing the accuracy of aircrafts’ taxi time prediction. The proposed MT-DSTGAN model is implemented and assessed through a case study of Beijing Capital International Airport with a real-world dataset. The advantage of the proposed model, which shows better performance in various evaluation metrics, is demonstrated in a comparative study with other baseline works. In summary, the proposed MT-DSTGAN exhibits promising capabilities in perceiving the dynamic changes in the taxiing process of aircraft and demonstrates the ability to capture complex spatio-temporal correlations in airport traffic data.
... In addition, the important features are extracted from the random forest model using backward exclusion, which is a quantitative analysis of the importance of features that is rarely seen in previous studies. Pham et al. [72] considered the controller's decision preferences and achieved high-precision taxi-out time prediction using both random forest regression and linear regression. In 2022, Zhang et al. [66] established a prediction model based on random forest regression and kernel density estimation, and used the kernel density estimation method to fit the set of results predicted by the decision tree with probability distributions, to obtain the probability density function of the taxi time; this method can analyze the probability distribution of the taxi time of a single aircraft while predicting the uncertainty of the taxi time. ...
... (1) Research on scientific feature extraction methods for taxi time prediction: Existing studies mainly consider input features such as surface arrival/departure flow rates, taxiing distance, pushback time, and departure queue length for predicting taxi time based on historical data. However, other factors such as surface congestion, human factors [72,78], and traffic management strategies that influence taxi time have received limited attention. Future studies should conduct in-depth analyses of factors affecting taxi time to determine the characteristic variables for the prediction model. ...
Airport arrival and departure movements are characterized by high dynamism, stochasticity, and uncertainty. Therefore, it is of paramount importance to predict and analyze surface taxi time accurately and scientifically. This paper conducts a comprehensive review of existing studies on surface taxi time prediction and analysis. Firstly, the overall research framework of surface taxi time prediction and analysis is categorized from three perspectives: taxi time type, movement type, and modeling method. Then, focusing on the two means of taxi time analytical modeling and simulation modeling, the existing mainstream models and methods are categorized, and the main ideas and scope of application of the various methods are analyzed. Finally, the paper presents the future development direction of surface taxi time prediction prospects. The research results are aimed at providing basic support and methodological guidance for reducing the uncertainty in airport surface operation and enhancing the level of control and decision-making ability of airport surface operation.
... Machine learning approaches are particularly suitable for capturing these subtle high-dimensional dependencies and have shown to provide more accurate predictions compared to conventional methods. 6,7) The implementation of such machine learning approaches is further supported by the availability of high-resolution surface movement data from Advanced Surface Movement Guidance and Control Systems (ASMGCS), which allows segmentation of the taxi process into different movement phases (e.g., taxiing over ramp, along taxiways, and in the departure queue). ...
This paper presents a machine learning-based approach for predicting the taxi-out time, with the departure process decomposed into two components: the time taken to travel from the gate to the departure queue, and the time spent in the departure queue. Gradient-Boosted Decision Tree (GBDT) models are trained to predict the two components using different feature sets, and a comparison of both model shows that they can provide better prediction accuracy compared with conventional methods, with a Root Mean Squared Error (RMSE) of 1.79 minutes and 0.92 minutes when predicting the taxiing and queuing times respectively, and 78% and 96% of predictions falling within a ±2 minute error margin. Predictions from the GBDT model are analysed and interpreted using SHAP (SHapley Additive exPlanations) values, a well-recognised technique for providing interpretability to many different black-box models, and allowing feature importance to be evaluated at global (model) and local (individual prediction) levels. In particular, the most important feature groups for the taxiing and queuing models are respectively the route features and runway queuing features. The model explainability provides a pathway towards the certification of machine learning techniques in Air Traffic Controller (ATCO) decision support tools.
Airport taxi delays adversely affect airports and airlines around the world leading to airside congestion, increased Air Traffic Controllers/Pilot workload, and adverse environmental impact due to excessive fuel burn. Airport Departure Metering (DM) is an effective approach to contain taxi delays by controlling departure pushback timings. The key idea behind DM is to transfer aircraft waiting time from taxiways to gates. State-of-the-art DM methods use model-based control policies that rely on airside departure modeling to obtain simplified analytical equations. Consequently, these models fail to capture non-stationarity in the airside operations leading to poor performance of control policies under uncertainties. This work proposes model-free and learning-based DM using Deep Reinforcement Learning (DRL) approach to reduce taxi delays while meeting flight schedule constraints. This paper casts the DM problem in a markov decision process framework and develops a representative airport-airside simulator to simulate airside operations and evaluate the learnt DM policy. For effective state representation, this work introduces taxiway hotspot features to account for the spatial-temporal evolution of airside congestion levels. This significantly improves the DM policy convergence rate during training. The performance of the learnt policy is evaluated under different traffic densities with a reduction of approximately 44% in taxi out delays, in medium-density traffic scenarios, which corresponds to 2-minute savings in taxi-out time per aircraft. Furthermore, benchmarking DRL against an evolutionary method and another state-of-the-art simulation-based heuristic demonstrates the superior performance of our method, especially in high traffic density scenarios. With increased traffic density, taxi-time savings achieved by the learnt DM policy increase without a significant decrease in runway throughput. Results, on a typical day of simulated operations at Singapore Changi Airport, demonstrate that DRL can learn an effective DM policy to contain congestion on the taxiways, reduce total fuel consumption by approximately 22% and better manage the airside traffic.
The key objective of this paper was to report on one of the industrial-based change case studies of the MASCA project (MAnaging System Change in Aviation—EU FP7, 2010–2013
). This case study provides a systematic insight into one airport’s approach to their preparation for full implementation of Airport Collaborative Decision Making (A-CDM). An action-based methodological approach was applied over a 3-year period, and a particular focus of this paper is on the application of the MASCA system change and operational evaluation tool (SCOPE/Structured Enquiry). Key recommendations resulted in research-led interventions, such as the development of a Serious Game to facilitate co-ordination and communications. The paper also reports on future recommendations for the implementation of A-CDM, such as prioritising social relations and trust building amongst airport stakeholders as opposed to viewing A-CDM solely as an IT-led project. Recommendations and learning from this case study can also be disseminated to other airports who are about to embark on the preparation for full A-CDM implementation and compliance.
Global Navigation Satellite Systems (GNSS) such as GPS and digital road maps can be used for land vehicle navigation systems. However, GPS requires a level of augmentation with other navigation sensors and systems such as Dead Reckoning (DR) devices, in order to achieve the required navigation performance (RNP) in some areas such as urban canyons, streets with dense tree cover, and tunnels. One of the common solutions is to integrate GPS with DR by employing a Kalman Filter (Zhao et al., 2003). The integrated navigation systems usually rely on various types of sensors. Even with very good sensor calibration and sensor fusion technologies, inaccuracies in the positioning sensors are often inevitable. There are also errors associated with spatial road network data. This paper develops an improved probabilistic Map Matching (MM) algorithm to reconcile inaccurate locational data with inaccurate digital road network data. The basic characteristics of the algorithm take into account the error sources associated with the positioning sensors, the historical trajectory of the vehicle, topological information on the road network (e.g., connectivity and orientation of links), and the heading and speed information of the vehicle. This then enables a precise identification of the correct link on which the vehicle is travelling. An optimal estimation technique to determine the vehicle position on the link has also been developed and is described. Positioning data was obtained from a comprehensive field test carried out in Central London. The algorithm was tested on a complex urban road network with a high resolution digital road map. The performance of the algorithm was found to be very good for different traffic maneuvers and a significant improvement over using just an integrated GPS/DR solution.
With the expected continued increases in air transportation, the mitigation of the consequent delays and environmental effects is becoming more and more important, requiring increasingly sophisticated approaches for airside airport operations. Improved on-stand time predictions (for improved resource allocation at the stands) and take-off time predictions (for improved airport-airspace coordination) both require more accurate taxi time predictions, as do the increasingly sophisticated ground movement models which are being developed. Calibrating such models requires historic data showing how long aircraft will actually take to move around the airport, but recorded data usually includes significant delays due to contention between aircraft. This research was motivated by the need to both predict taxi times and to quantify and eliminate the effects of airport load from historic taxi time data, since delays and re-routing are usually explicitly considered in ground movement models. A prediction model is presented here that combines both airport layout and historic taxi time information within a multiple linear regression analysis, identifying the most relevant factors affecting the variability of taxi times for both arrivals and departures. The promising results for two different European hub airports are compared against previous results for US airports.
Third-generation personal navigation assistants (PNAs) (i.e., those that provide a map, the user's current location, and directions) must be able to reconcile the user's location with the underlying map. This process is known as map matching. Most existing research has focused on map matching when both the user's location and the map are known with a high degree of accuracy. However, there are many situations in which this is unlikely to be the case. Hence, this paper considers map matching algorithms that can be used to reconcile inaccurate locational data with an inaccurate map/network.
Taxi routing for aircraft: Creation and controlling
- I Gerdes
- A Temme
I. Gerdes and A. Temme, "Taxi routing for aircraft:
Creation and controlling," The Second SESAR
Innovation Days, 2012.
A density-based algorithm for discovering clusters in large spatial databases with noise
- M Ester
- H.-P Kriegel
- J Sander
- X Xu
M. Ester, H.-P. Kriegel, J. Sander, X. Xuet al., "A
density-based algorithm for discovering clusters in
large spatial databases with noise." In Kdd, vol. 96, no.
34, 1996, pp. 226-231.