Mostra el registre d'ítem simple
Cost index (CI) and take-off mass (TOM) estimation using machine learning algorithms
dc.contributor | Dalmau Codina, Ramon |
dc.contributor | Prats Menéndez, Xavier |
dc.contributor.author | Gil Vidal, Santiago |
dc.contributor.author | Olivares Guixé, Carles |
dc.date.accessioned | 2017-07-26T11:19:01Z |
dc.date.available | 2017-07-26T11:19:01Z |
dc.date.issued | 2017-06-22 |
dc.identifier.uri | http://hdl.handle.net/2117/106860 |
dc.description.abstract | The Cost Index (CI) and Take-off Mass (TOM) are two parameters that are very important in order to study the preferences on airlines operation. In the same way, these two parameters would allow to predict ground-based trajectories accurately. Nowadays, unfortunately, this information is not shared by the airlines, because this information is confidential as they help to define market strategies of the airline. The objective of this final degree project is to develop and evaluate an algorithm able to estimate CI and TOM from data of a flight trajectory, that could be collected by a conventional antenna (i.e. radar data or ADS-B), by using Machine Learning algorithms. The algorithm should be trained with data from the PEP (Performance Program Airbus). The data will be shaped by thousands of trajectories generated with different ranges of distances, TOM, CI and atmospheric conditions in order to establish the input training data for Machine Learning. Once the algorithm has been generated, to ensure its robustness, it will be tested with data containing noise where the influence of the parameters in the prediction would be evaluated. Finally, it will be validated with new aircraft trajectories from PEP. The ultimate goal of the final degree project is to check and perform the study with real flight data. To realize this, radar data will be obtained from the DDR2 platform of Eurocontrol. With some flights trajectories, we will study the values of CI and TOM used by several airlines with the Machine Learning algorithm previously trained. In conclusion, it has been demonstrated that for CI the most relevant input variable is the Mach Number because it is the most visible evidence given to the time-fuel cost relation. On the other hand, TOM is more related to the distance of the flight and flight levels (FL). When the prediction algorithm is applied to real cases flights, we observed that low-cost airlines and flag carriers use different strategies of CI. Even so, a single airline usually use the same CI for most of their routes, wasting the opportunity to optimize the costs of the route and all the advantages offered by the CI. |
dc.language.iso | eng |
dc.publisher | Universitat Politècnica de Catalunya |
dc.rights.uri | http://creativecommons.org/licenses/by-nc-sa/3.0/es/ |
dc.subject | Àrees temàtiques de la UPC::Aeronàutica i espai |
dc.subject.lcsh | Airlines--Management |
dc.subject.other | Machine learning |
dc.subject.other | Cost index |
dc.subject.other | Parameters estimation |
dc.title | Cost index (CI) and take-off mass (TOM) estimation using machine learning algorithms |
dc.type | Bachelor thesis |
dc.subject.lemac | Línies aèries -- Costos |
dc.rights.access | Open Access |
dc.date.updated | 2017-07-01T04:24:16Z |
dc.audience.educationlevel | Estudis de primer/segon cicle |
dc.audience.mediator | Escola d'Enginyeria de Telecomunicació i Aeroespacial de Castelldefels |
dc.audience.degree | GRAU EN ENGINYERIA D'AEROPORTS (Pla 2010) |