Please use this identifier to cite or link to this item: http://hdl.handle.net/10609/110146
Title: Avaluació de models basats en intel·ligència artificial per a la predicció espacial del risc d'incendi forestal
Author: Bonet-Vilela, Fidel  
Director: Ventura, Carles  
Tutor: Kanaan-Izquierdo, Samir  
Abstract: The goal of this final master's project is to obtain spatial prediction models of forest fire risk through the combined use of artificial intelligence, geographic information systems and big data. Several sets of fire, meteorological, orographic and vegetation data have been used to predict fire risk areas and to estimate the probability by means of several machine learning techniques: classification to predict the risk of forest fire, regression to predict the size of the fires at the moment of ignition and clustering to obtain fire risk areas according to the meteorological conditions. Robust models have been obtained with accuracies of up to 90% in predicting risk and 99% in grouping new examples into weather-dependent risk categories. The best results have been obtained with the use of deep learning. Specifically, genetic algorithms have been used to optimize the architecture of a multilayer perceptron. Finally, the results of the project allow us to obtain risk maps with sufficient detail for various areas (counties, municipalities, natural spaces, etc.) and valid for specific areas such as a natural park where the results achieved have allowed us to estimate the wildfire risk for the various areas of the park and even in certain sensitive places such as the main paths and vehicle parking areas.
Keywords: machine learning
geographic information systems
big data
climate change
forest fires
Document type: info:eu-repo/semantics/masterThesis
Issue Date: 31-Dec-2019
Publication license: http://creativecommons.org/licenses/by-nc-nd/3.0/es/  
Appears in Collections:Bachelor thesis, research projects, etc.

Files in This Item:
File Description SizeFormat 
fbonetviTFM0120memòria.pdfMemòria del TFM33,52 MBAdobe PDFThumbnail
View/Open
fbonetviTFM0120presentació.pdfPresentació del TFM40,29 MBAdobe PDFThumbnail
View/Open