Forest fires are a severe and challenging environmental emergency, which threaten ecosystems and infrastructure. To predict the behaviour of their spread, computational tools can be applied. These forest fire simulators require a multitude of environmental parameters as input, many of them outdated at prediction time or imprecise or missing. To decrease the effect of parameter uncertainty on the fire simulation outcome, parameter calibration methods can be applied.
This thesis describes a parameter calibration framework based on Evolutionary Intelligent Systems. A Genetic Algorithm is coupled with an intelligent paradigm to speed up the usually very time-consuming calibration process. Parameter calibration can thus be enabled in case of large forest fires or limited computational resources, while maintaining prediction quality.
© 2001-2024 Fundación Dialnet · Todos los derechos reservados