Título: The Levenberg-Marquardt Method Applied in Nonlinear Dynamic Systems
Autores: Silva;A. P. C.;Khater, E.
Resumo: This work consist in the identification of nonlinear dynamic systems using two functions with the backpropagation algorithm. Four optimization methods were implemented for a convergence comparation: Gradient, Momentum, Conjugate Gradient and Levenberg-Marquardt. The main objective was show, by the results, that the Levenberg-Marquardt method presented a better performance in the identification of the researched systems. Moreover, this work investigated the influence of the inputs in the network training performance.
Código DOI: 10.21528/CBRN2001-116
Artigo em pdf: 5cbrn_116.pdf
Arquivo BibTex: 5cbrn_116.bib