Increasing trajectory prediction accuracy under bad ... 2015 - ASDA... · Increasing trajectory...
Transcript of Increasing trajectory prediction accuracy under bad ... 2015 - ASDA... · Increasing trajectory...
![Page 1: Increasing trajectory prediction accuracy under bad ... 2015 - ASDA... · Increasing trajectory prediction accuracy under bad weather conditions Norbert Fuerstenau (DLR) & Guglielmo](https://reader030.fdocuments.us/reader030/viewer/2022040310/5d4dde1c88c993a01f8b520a/html5/thumbnails/1.jpg)
Increasing trajectory prediction accuracy under bad weather conditions
Norbert Fuerstenau (DLR)&
Guglielmo Lulli (University of Lancaster)
![Page 2: Increasing trajectory prediction accuracy under bad ... 2015 - ASDA... · Increasing trajectory prediction accuracy under bad weather conditions Norbert Fuerstenau (DLR) & Guglielmo](https://reader030.fdocuments.us/reader030/viewer/2022040310/5d4dde1c88c993a01f8b520a/html5/thumbnails/2.jpg)
Sub-challenge 1: approaches for uncertainty management in trajectory prediction
![Page 3: Increasing trajectory prediction accuracy under bad ... 2015 - ASDA... · Increasing trajectory prediction accuracy under bad weather conditions Norbert Fuerstenau (DLR) & Guglielmo](https://reader030.fdocuments.us/reader030/viewer/2022040310/5d4dde1c88c993a01f8b520a/html5/thumbnails/3.jpg)
• Characterization of uncertainty
Sub-challenge 1: approaches for uncertainty management in trajectory prediction
1-d 2-d 3-d
• Large-Disruptive (weather phenomenon, other disruption, etc. )• Normal Traffic uncertainty (inherent stochasticity of the system)
![Page 4: Increasing trajectory prediction accuracy under bad ... 2015 - ASDA... · Increasing trajectory prediction accuracy under bad weather conditions Norbert Fuerstenau (DLR) & Guglielmo](https://reader030.fdocuments.us/reader030/viewer/2022040310/5d4dde1c88c993a01f8b520a/html5/thumbnails/4.jpg)
Optimization under uncertainty, e.g., stochastic programming, robust optimization, etc.
RO’94,VBO’94, AEO’00, HM’07, BHM’10, ADL’11, AAEP’12, BS15, CLNV’16, BC’16
“Macro” uncertainty
]),([)( 1121111
xfExcMinXx
Computational challenges:• Multi-dimensional integration• Integer Requirement• Large volume of data
MinxX
f (x) : g(x,u) b uU
![Page 5: Increasing trajectory prediction accuracy under bad ... 2015 - ASDA... · Increasing trajectory prediction accuracy under bad weather conditions Norbert Fuerstenau (DLR) & Guglielmo](https://reader030.fdocuments.us/reader030/viewer/2022040310/5d4dde1c88c993a01f8b520a/html5/thumbnails/5.jpg)
Methods to measure and forecast congestion at airports K’76, M’95,PBO’95, WFM’04, H et al.’09, GNS’11, NH’12
“Micro” uncertainty
![Page 6: Increasing trajectory prediction accuracy under bad ... 2015 - ASDA... · Increasing trajectory prediction accuracy under bad weather conditions Norbert Fuerstenau (DLR) & Guglielmo](https://reader030.fdocuments.us/reader030/viewer/2022040310/5d4dde1c88c993a01f8b520a/html5/thumbnails/6.jpg)
Optimized Approach Sequencing (WP-E Robust-ATM, FAU-DLR, 2013-16)
Empirical Airport /A-SMGCSTraffic Data : Arrival / Departure Delays with -PDF Model
Single 17h-Day: 209 Arrivals ai = 0 min
Single Flight 176 Departures: 7-12/2013 = 18 min
Monte-Carlo Simulation of pre-tactical Arrival Scheduling with Random Departure-Delays
Baseline:FCFS, Arrival Delays
FAUOptimizer(discrete,Gurobi-nominal)
[ET, STA, LT, Ltmax]
Dpt.-Delay -PDF(a,)
![Page 7: Increasing trajectory prediction accuracy under bad ... 2015 - ASDA... · Increasing trajectory prediction accuracy under bad weather conditions Norbert Fuerstenau (DLR) & Guglielmo](https://reader030.fdocuments.us/reader030/viewer/2022040310/5d4dde1c88c993a01f8b520a/html5/thumbnails/7.jpg)
Combining Optimized Sequencing with Dynamic Performance Predictionunder Disruptive Weather. Example Low Pressure Event „Christian“
> SID Research Tournament> Norbert Fürstenau• Document > 17/11/2015DLR.de • Chart 7
Wind speed & Gust data (METAR)
Fit with Dynamical Systems
Disturbance Model: Speed v(t)
Dynamical Systems Model forDisturbance & Performance prediction
optimized Management Actions G(t)
Arrival Rate Data & Fit withDynamical Systems Model forPerformance-Disruption and Recovery
Anticipating Actions Recovery Actions
![Page 8: Increasing trajectory prediction accuracy under bad ... 2015 - ASDA... · Increasing trajectory prediction accuracy under bad weather conditions Norbert Fuerstenau (DLR) & Guglielmo](https://reader030.fdocuments.us/reader030/viewer/2022040310/5d4dde1c88c993a01f8b520a/html5/thumbnails/8.jpg)
Case of Heathrow airport*
* Caccavale, M.V., Iovanella, A., Lancia, C., Lulli, G., Scoppola, B. 2014. A model of inbound air traffic: the application to Heathrow airport. Journal of Air Transport Management 34, p. 116‐122.
![Page 9: Increasing trajectory prediction accuracy under bad ... 2015 - ASDA... · Increasing trajectory prediction accuracy under bad weather conditions Norbert Fuerstenau (DLR) & Guglielmo](https://reader030.fdocuments.us/reader030/viewer/2022040310/5d4dde1c88c993a01f8b520a/html5/thumbnails/9.jpg)
Sub-challenge 2: identifying the need for prediction accuracy and defining the required level of certainty.
Iovanella, Scoppola, Pozzi, Tedeschi: The Impact of 4D Trajectories on ArrivalDelays in Mixed Traffic Scenarios, SID 2011.
![Page 10: Increasing trajectory prediction accuracy under bad ... 2015 - ASDA... · Increasing trajectory prediction accuracy under bad weather conditions Norbert Fuerstenau (DLR) & Guglielmo](https://reader030.fdocuments.us/reader030/viewer/2022040310/5d4dde1c88c993a01f8b520a/html5/thumbnails/10.jpg)
Sub-challenge 2: identifying the need for prediction accuracy and defining the required level of certainty.
Iovanella, Scoppola, Pozzi, Tedeschi: The Impact of 4D Trajectories on ArrivalDelays in Mixed Traffic Scenarios, SID 2011.
![Page 11: Increasing trajectory prediction accuracy under bad ... 2015 - ASDA... · Increasing trajectory prediction accuracy under bad weather conditions Norbert Fuerstenau (DLR) & Guglielmo](https://reader030.fdocuments.us/reader030/viewer/2022040310/5d4dde1c88c993a01f8b520a/html5/thumbnails/11.jpg)
Sub-challenge 2: identifying the need for prediction accuracy and defining the required level of certainty.• Is it enough?
![Page 12: Increasing trajectory prediction accuracy under bad ... 2015 - ASDA... · Increasing trajectory prediction accuracy under bad weather conditions Norbert Fuerstenau (DLR) & Guglielmo](https://reader030.fdocuments.us/reader030/viewer/2022040310/5d4dde1c88c993a01f8b520a/html5/thumbnails/12.jpg)
• How to absorb time in a time constrained environment in the most efficient manner is a key issue, e.g., reducing speed in cruise (K. et al ’11, DP’12).
• Aircraft operators behaviour Gap between system efficiency and individual efficiency Rules / decision mechanisms / incentives to follow the “advised decisions”
Experience
![Page 13: Increasing trajectory prediction accuracy under bad ... 2015 - ASDA... · Increasing trajectory prediction accuracy under bad weather conditions Norbert Fuerstenau (DLR) & Guglielmo](https://reader030.fdocuments.us/reader030/viewer/2022040310/5d4dde1c88c993a01f8b520a/html5/thumbnails/13.jpg)
![Page 14: Increasing trajectory prediction accuracy under bad ... 2015 - ASDA... · Increasing trajectory prediction accuracy under bad weather conditions Norbert Fuerstenau (DLR) & Guglielmo](https://reader030.fdocuments.us/reader030/viewer/2022040310/5d4dde1c88c993a01f8b520a/html5/thumbnails/14.jpg)
> Lecture > Author • Document > DateDLR.de • Chart 14