CBL - Campus del Baix Llobregat

Projecte llegit

Títol: Predicción de condiciones meteorológicas en aeropuertos mediante deep learning

Estudiants que han llegit aquest projecte:

Director/a: TARRÉS RUIZ, FRANCESC

Departament: TSC

Títol: Predicción de condiciones meteorológicas en aeropuertos mediante deep learning

Data inici oferta: 23-06-2022     Data finalització oferta: 23-02-2023


Estudis d'assignació del projecte:
    DG ENG AERO/TELEMÀT
Tipus: Individual
 
Lloc de realització: EETAC
 
Paraules clau:
inteligencia artificial, meteorología, deep learning
 
Descripció del contingut i pla d'activitats:
El proyecto consiste en utilizar bases de datos históricas sobre condiciones

metereológicas en aeropuertos para diseñar un modelo de predicción basado en redes

neuronales y deep learning. El estudiante deberá estudiar las bases de datos disponibles

para analizar los parámetros disponibles y seleccionar los parámetros a predecir. Estas

bases de datos deberán reestructurarse en un entorno Mongo para facilitar el acceso a

los datos a través de Python para el entrenamiento del modelo de predicción y para la

realización de inferencias. Se evaluaran e implementaran diferentes modelos de redes

neuronales para realizar la estimación, valorando los resultados obtenidos y

seleccionando aquellos que presenten mejores resultados. Finalmente, con los modelos

seleccionados se realizará un aplicativo que pueda usarse para realizar proyecciones de

las condiciones metereologicas a diferentes intervalos temporales (2 horas, 4 horas, 12

horas).



 
Overview (resum en anglès):
It is indisputable that Artificial Intelligence has been progressing meaningfully in recent years. In particular, the branch of Deep Learning, on which this project will focus.
This technology is present in applications such as automatic image classification, speech recognition, autonomous driving, visual or analytical predictions, etc. In this particular case, in weather forecasting.
On the other hand, in the airport environment, it is essential to have control over the weather and possible momentary changes that may occur, as they can be critical in aeronautical operations. To achieve this, daily meteorological data will be obtained from the AEMET database, and Artificial Intelligence will be trained using this data to predict wind, precipitation, and visibility variables for the next two to four hours.
Therefore, the objective of this project is to present a tool that has real utility in airports. This application will combine the aforementioned elements to create a weather prediction application for selected airports using different Deep Learning models. The effectiveness and performance of each chosen method will be analyzed, and potential improvements will be discussed.
The work begins with an introduction, where the study objectives, the intended process, and possible difficulties are presented. Following that, Chapter 1 discusses the evolution of Artificial Intelligence, Chapter 2 presents the theoretical foundations of neural networks and meteorology. Chapter 3 develops the methodology to be followed throughout the project, and Chapter 4 applies it using Python libraries.
In Chapter 5, the differences and limitations of each model results are discussed. Finally, Chapter 6 concludes various points observed throughout this work and discusses future potential improvements and research in this field.

© CBLTIC Campus del Baix Llobregat - UPC