Publicar

55 - DESENVOLVIMENTO DE FERRAMENTA DE APOIO À PREVENÇÃO DE DESLIZAMENTOS COM USO DE MACHINE LEARNING E INTEGRAÇÃO DE DADOS GEOESPACIAIS

Doi: 10.47094/COBRAMSEG2026.4/55

PALAVRAS CHAVE: Georreferenciamento, Aprendizado de máquina, Geotecnia, Programação.

KEYWORDS: Georeferencing, Machine learning, Geotechnics, Programming.

ABSTRACT:
This work presents the preliminary results from an ongoing project, carried out in partnership between the Federal University of Bahia and the Civil Defense of Salvador (CODESAL), with the objective of supporting preventive actions in urban slope areas. A computational tool was developed in Python to identify areas susceptible to slope instability, based on Michalowski's nomograms and an integrated set of georeferenced data, including a Digital Terrain Model (DTM), mechanical properties of the main soil formations in the region, and relevant geological structures. This data feeds machine learning algorithms responsible for estimating the probability of landslides. Preliminary results indicate good performance of the tool, with high predictive capacity in different contexts of the urban area. The methodology adopted allows for continuous updating of the database, contributing to the improvement of predictions and risk mitigation, with the potential to make preventive interventions more assertive.

Autores

  • Guilherme Barbosa de Santana

  • Jailma Santos de Souza de Oliveira

  • Miriam de Fátima Carvalho

  • Sandro Lemos Machado

  • Élio Gões Perrone Júnior

Download