Identificação e linearização por redes neurais de um manipulador robotico com acoplamento flexivel




The goal of this work is to present a plan of manipulators feedback linearization using neural networks based on adaptive control techniques. The behavior of designed robot will be studied through a system with flexible joint. This manipulator is composed by three DOF (degrees of freedom), with two DOF in rigid segments of rotation motion and translation motion, and one flexible joint moved for a "toothed belt" or "timing belt" between two rotary shaft, forming the third DOF. Initially it will be presented the dynamic feedback linearization method. Then artificial neural networks of multiple layers will be used in order to replace the blocks of linearization and planto This analysis inc1ude the possibility to work with unknown parameters, that can be considered as an adaptive method to the system


sistemas lineares redes neurais (computação) identificação de sistemas robotica

Documentos Relacionados