Título: Recuperabilidade de Automóveis pela Análise da Coluna da Carroçaria através de Redes Neurais Artificiais
Autores: Xanthopulo, Walter; Martins, Weber
Resumo: This paper describes a methodological study for applying technology based on neural networks to the assessment and classification of deformations in crashed cars. Generally, this study applies to every material that obeys the Hook Law. From this study, a procedure could be developed by the computation of deviations of car columns with respect to the standard (brand new) one in order to make it possible to infer the recuperability state of the car. Some simulation results (using Multilayer Perceptrons with backpropagation) are shown and clearly demonstrate that neural networks are a promising technique to cope with similar situations.
Código DOI: 10.21528/CBRN2001-081
Artigo em pdf: 5cbrn_081.pdf
Arquivo BibTex: 5cbrn_081.bib