Ajuste evolutivo de parâmetros de autômatos celulares probabilísticos em modelos de propagação de incêndios

• Autor(a) principal: Ferreira, Maria Eugênia de Ávila
• Data de Publicação: 2023
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da UFU
• Texto Completo: https://repositorio.ufu.br/handle/123456789/38738; http://doi.org/10.14393/ufu.di.2023.192
• topic: Algoritmos genéticos; Genetic algorithms; Autômato celular; Cellular automata; Modelo de propagação de incêndios; Fire propagation model; Computação evolutiva; Evolutionary computing; Ajustes de parâmetros; Parameters adjustments; Computação; Computing; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO; Computação; Automato celular; Algorítmos genéticos; Programação evolutiva (Computação)

Forest fires cause diverse damages, whether environmental or socio-environmental. They have significantly increased as a result of climate change, which affects various biomes. In order to assist experts in containing these phenomena, the spread of a forest fire can be simulated through computational models. Modeling the spread of fire is essential in preventing and controlling the damage caused by fires in areas with native vegetation. Cellular automata (CA) are discrete models that represent a lattice of cells that interact with each other. Their use has been investigated in modeling various natural phenomena, including the spread of fires. However, for satisfactory modeling, characteristics of the biome such as vegetation type, soil, climate, wind, and terrain topography should be considered. Adjusting the various parameters involved in this modeling can be a challenging and complex task. In this context, this study proposes an evolutionary method based on Genetic Algorithms to adjust the parameters of fire spread models based on probabilistic cellular automata rules. In order to evaluate the effectiveness of the proposed approach and its sensitivity to variations in the data to be modeled, several experiments were carried out, considering scenarios with homogeneous and heterogeneous vegetation and different biome characteristics. In this process, due to the lack of real fire data, data generated from CA and two other models available in the literature were employed. The results achieved show that the evolutionary algorithm is capable of automatically adjusting the parameters of a fire spread model.