Una comparativa entre redes neuronales artificiales y métodos clásicos para la predicción de la movilidad entre zonas de transporte. Aplicación práctica en el Campo de Gibraltar, Espa?a

摘要

Traffic issues are more common every day due to the great technological development of humanity. Therefore, the control is essential to optimize infrastructure and public transport. To achieve this goal, it is necessary to make an estimate of the demand of the mobility. An alternative method, based on Artificial Neural Networks (ANNs), has been analyzed in this work comparing to traditional prediction techniques. The aim is to obtain an estimation procedure using simple, economical input variables which are easy to find. Unlike traditional models. These new models are able to perform a better fitting of input-output mapping. The results are encouraging and therefore the ability of ANNs is shown to estimate mobility between zones.