Earthquake risk induction models with genetic algorithm

Detalhes bibliográficos
Autor(a) principal: Lavinas, Yuri Cossich
Data de Publicação: 2016
Tipo de documento: Trabalho de conclusão de curso
Idioma: por
Título da fonte: Biblioteca Digital de Monografias da UnB
Texto Completo: http://bdm.unb.br/handle/10483/14084
Resumo: Monografia (graduação)—Universidade de Brasília, Instituto de Ciências Exatas, Departamento de Ciência da Computação, 2016.
id UNB-2_e2ab7423edc4aad23d83ac1b19035fad
oai_identifier_str oai:bdm.unb.br:10483/14084
network_acronym_str UNB-2
network_name_str Biblioteca Digital de Monografias da UnB
repository_id_str 11571
spelling Lavinas, Yuri CossichAranha, Diego de FreitasLadeira, MarceloLAVINAS, Yuri Cossich. Earthquake risk induction models with genetic algorithm. 2016. x, 54 f., il. Monografia (Bacharelado em Ciência da Computação)—Universidade de Brasília, Brasília, 2016.http://bdm.unb.br/handle/10483/14084Monografia (graduação)—Universidade de Brasília, Instituto de Ciências Exatas, Departamento de Ciência da Computação, 2016.Este projeto visa desenvolver um modelo de previsão de riscos de terremotos com Algoritmos Geneticos (GA). Modelos de risco de terremotos descrevem o risco de ocorrência de atividades sísmicas em uma determinada área baseado em informações previamente obtidas de terremotos em regiões próximas da área de estudo. GA foi utilizada para aprender um modelo de risco usando informações previamente obtidas como base de treino. Baseado nos resultados obtidos, acreditamos ser possível obter melhores modelos se conhecimentos do domínio da aplicação, como conhecimentos oriundos da literatura ou modelos de distribuição de terremotos, poderem ser incorporados ao processo de aprendizado do Algoritmo Genético. O objetivo principal é definir um método para estimar a probabilidade de ocorrências de terremotos no Japão usando dados históricos de terremotos para um grupo de determinadas regiões geográficas. Este trabalho se baseia no contexto do “Collaboratory for the Study of Earthquake Predictability” (CSEP), que visa padronizar os estudos e testes de modelos de previsão de riscos de terremotos. Durante o desenvolvimento das atividades, passamos por três estágios. (1) Nós propusemos um método baseado em uma aplicação de GA e objetivamos gerar um método estatístico de análise de risco de terremotos. Estes foram analisados por seus valores de log-likelihood, como sugerido pelo Regional Earthquake Likelihood Model (RELM). (2) A seguir, modificiamos a representação do genoma, de uma representação baseada em área para uma representação baseada em ocorrências de terremotos, buscando obter uma convergência mais rápida dos valores de log-likelihood dos candidatos do GA e (3) usamos métodos da sismologia conhecidos para refinar os candidatos gerados pelo GA. Em todas as estapas, os modelos de risco são comparados com dados reais, com modelos gerados pela aplicação do Relative Intensity Algorithm (RI) e com eles próprios. Os dados utilizados foram obtidos pela Japan Metereological Angency (JMA) e são relativos a atividades de terremotos no Japão entre os anos de 2000 e 2013. Nós analisamos as contruibuições de cada modelo proposto usando metodologia descritas pelo CSEP e comparamos os modelos desenvolvidos. Os resultados apontam que modelos com terremotos mais estáveis possuem maiores valores de log-likelihood.Submitted by Nayara Silva (nayarasilva@bce.unb.br) on 2016-08-05T17:32:34Z No. of bitstreams: 3 license_text: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) 2016_YuriCossichLavinas.pdf: 2344440 bytes, checksum: 6c68745c613eed37a0d0481328117d0e (MD5)Approved for entry into archive by Ruthlea Nascimento (ruthlea.nascimento@gmail.com) on 2016-08-11T19:14:56Z (GMT) No. of bitstreams: 3 license_text: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) 2016_YuriCossichLavinas.pdf: 2344440 bytes, checksum: 6c68745c613eed37a0d0481328117d0e (MD5)Made available in DSpace on 2016-08-11T19:14:56Z (GMT). No. of bitstreams: 3 license_text: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) 2016_YuriCossichLavinas.pdf: 2344440 bytes, checksum: 6c68745c613eed37a0d0481328117d0e (MD5)This projects aims to develop an earthquake prevision risk model using Genetic Algorithms (GA). Earthquake Risk Models describe the risk of occurrence of seismic events on a given area based on information such as past earthquakes in nearby regions, and the seismic properties of the area under study. We used GA to learn risk models using past earthquake occurrence as training data. Based on the results obtained, we believe that a much better model could be learned if domain knowledge, such as known theories and models on earthquake distribution, were incorporated into the Genetic Algorithm’s training process. The main goal is to define good methods to estimate the probability of earthquake occurrences in Japan using historical data of a group given geographical regions. This work is established in the context of the “Collaboratory for the Study of Earthquake Predictability” (CSEP), which seeks to standardise the studies and tests of earthquake risk prevision models. To achieve the main goal, we passed three stages. (1) We proposed a method based in one application of GA and aims to develop statistical methods of analysis of earthquake risk. The risk models generated by this application were analysed by their log-likelihood values, as suggested by the Regional Earthquake Likelihood Model (RELM). (2) Then, we modify the genome representation from an area-based representation to an earthquake representation aiming to reach a faster convergence of the log-likelihood values of the GA’s candidates and (3) we use known methods from seismology (such as the Omori- Utsu formula) to refine the candidates generated by the GA. In all stages, the risk models are compared with real data, with the models generated by the application of the Relative Intensity Algorithm (RI) and with themselves. The data used was obtained from the Japan Meteorological Agency (JMA) and are related with earthquake activity in Japan between the years of 2000 and 2013. We analyse the contributions from each risk model using the methodologies described in the CSEP and compare their quality. Our results indicate that models with more stable earthquakes obtain higher log-likelihood values.Earthquake risk induction models with genetic algorithminfo:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/bachelorThesis2016-08-11T19:14:56Z2016-08-11T19:14:56Z2016Algoritmos genéticosTerremotosinfo:eu-repo/semantics/openAccessporreponame:Biblioteca Digital de Monografias da UnBinstname:Universidade de Brasília (UnB)instacron:UNBORIGINAL2016_YuriCossichLavinas.pdf2016_YuriCossichLavinas.pdfapplication/pdf2344440http://bdm.unb.br/xmlui/bitstream/10483/14084/1/2016_YuriCossichLavinas.pdf6c68745c613eed37a0d0481328117d0eMD51CC-LICENSElicense_urllicense_urltext/plain49http://bdm.unb.br/xmlui/bitstream/10483/14084/2/license_url4afdbb8c545fd630ea7db775da747b2fMD52license_textlicense_textapplication/octet-stream0http://bdm.unb.br/xmlui/bitstream/10483/14084/3/license_textd41d8cd98f00b204e9800998ecf8427eMD53license_rdflicense_rdfapplication/octet-stream0http://bdm.unb.br/xmlui/bitstream/10483/14084/4/license_rdfd41d8cd98f00b204e9800998ecf8427eMD54LICENSElicense.txtlicense.txttext/plain1758http://bdm.unb.br/xmlui/bitstream/10483/14084/5/license.txt48fee5d355e169b5219b5efc5a9ad174MD5510483/140842016-08-11 16:14:56.931oai:bdm.unb.br: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 Digital de Monografiahttps://bdm.unb.br/PUBhttp://bdm.unb.br/oai/requestbdm@bce.unb.br||patricia@bce.unb.bropendoar:115712016-08-11T19:14:56Biblioteca Digital de Monografias da UnB - Universidade de Brasília (UnB)false
dc.title.pt_BR.fl_str_mv Earthquake risk induction models with genetic algorithm
title Earthquake risk induction models with genetic algorithm
spellingShingle Earthquake risk induction models with genetic algorithm
Lavinas, Yuri Cossich
Algoritmos genéticos
Terremotos
title_short Earthquake risk induction models with genetic algorithm
title_full Earthquake risk induction models with genetic algorithm
title_fullStr Earthquake risk induction models with genetic algorithm
title_full_unstemmed Earthquake risk induction models with genetic algorithm
title_sort Earthquake risk induction models with genetic algorithm
author Lavinas, Yuri Cossich
author_facet Lavinas, Yuri Cossich
author_role author
dc.contributor.advisorco.none.fl_str_mv Aranha, Diego de Freitas
dc.contributor.author.fl_str_mv Lavinas, Yuri Cossich
dc.contributor.advisor1.fl_str_mv Ladeira, Marcelo
contributor_str_mv Ladeira, Marcelo
dc.subject.por.fl_str_mv Algoritmos genéticos
Terremotos
topic Algoritmos genéticos
Terremotos
description Monografia (graduação)—Universidade de Brasília, Instituto de Ciências Exatas, Departamento de Ciência da Computação, 2016.
publishDate 2016
dc.date.submitted.none.fl_str_mv 2016
dc.date.accessioned.fl_str_mv 2016-08-11T19:14:56Z
dc.date.available.fl_str_mv 2016-08-11T19:14:56Z
dc.type.status.fl_str_mv info:eu-repo/semantics/publishedVersion
dc.type.driver.fl_str_mv info:eu-repo/semantics/bachelorThesis
format bachelorThesis
status_str publishedVersion
dc.identifier.citation.fl_str_mv LAVINAS, Yuri Cossich. Earthquake risk induction models with genetic algorithm. 2016. x, 54 f., il. Monografia (Bacharelado em Ciência da Computação)—Universidade de Brasília, Brasília, 2016.
dc.identifier.uri.fl_str_mv http://bdm.unb.br/handle/10483/14084
identifier_str_mv LAVINAS, Yuri Cossich. Earthquake risk induction models with genetic algorithm. 2016. x, 54 f., il. Monografia (Bacharelado em Ciência da Computação)—Universidade de Brasília, Brasília, 2016.
url http://bdm.unb.br/handle/10483/14084
dc.language.iso.fl_str_mv por
language por
dc.rights.driver.fl_str_mv info:eu-repo/semantics/openAccess
eu_rights_str_mv openAccess
dc.source.none.fl_str_mv reponame:Biblioteca Digital de Monografias da UnB
instname:Universidade de Brasília (UnB)
instacron:UNB
instname_str Universidade de Brasília (UnB)
instacron_str UNB
institution UNB
reponame_str Biblioteca Digital de Monografias da UnB
collection Biblioteca Digital de Monografias da UnB
bitstream.url.fl_str_mv http://bdm.unb.br/xmlui/bitstream/10483/14084/1/2016_YuriCossichLavinas.pdf
http://bdm.unb.br/xmlui/bitstream/10483/14084/2/license_url
http://bdm.unb.br/xmlui/bitstream/10483/14084/3/license_text
http://bdm.unb.br/xmlui/bitstream/10483/14084/4/license_rdf
http://bdm.unb.br/xmlui/bitstream/10483/14084/5/license.txt
bitstream.checksum.fl_str_mv 6c68745c613eed37a0d0481328117d0e
4afdbb8c545fd630ea7db775da747b2f
d41d8cd98f00b204e9800998ecf8427e
d41d8cd98f00b204e9800998ecf8427e
48fee5d355e169b5219b5efc5a9ad174
bitstream.checksumAlgorithm.fl_str_mv MD5
MD5
MD5
MD5
MD5
repository.name.fl_str_mv Biblioteca Digital de Monografias da UnB - Universidade de Brasília (UnB)
repository.mail.fl_str_mv bdm@bce.unb.br||patricia@bce.unb.br
_version_ 1801493001963307008

Registros relacionados