Título: Controle Adaptativo de Estruturas Flexíveis via Redes Neurais
Autores: Gervini, Vitor I.; Hemerly, Elder M.; Gomes, Sebastião C. P.
Resumo: Active control law development for flexible structures is still an open problem, mainly due to nonlinear friction in the actuators. In the majority of actuators exists a significant nonlinear friction causing a deadzone in torque. The aim of this paper is to propose a simple on-line control law, using neural networks (NN), to compensate this nonlinear friction. The NN on-line training adds an adaptive feature to the compensating mechanism. We illustrate the controller effectiveness via simulations with a realistic model of a flexible structure, that was been modeled using a lumped mass approach. The results highlights the efficiency of the proposed mechanism to compensate the nonlinear friction torque, thereby enabling the vibrations attenuation of the flexible link, as well as, the tracking of small reference signals.
Código DOI: 10.21528/CBRN2003-083
Artigo em PDF: 6CBRN_083.PDF
Arquivo BibTex: 6CBRN_083.bib