Submitted by Automa??o e Estat?stica (sst@bczm.ufrn.br) on 2017-02-21T00:17:08Z No. of bitstreams: 1 BrennaKarolynaDosSantosSilva_DISSERT.pdf: 6981280 bytes, checksum: 0266d5d98ab38bb268fe5ddd9e7e3c97 (MD5) / Approved for entry into archive by Arlan Eloi Leite Silva (eloihistoriador@yahoo.com.br) on 2017-03-03T22:08:40Z (GMT) No. of bitstreams: 1 BrennaKarolynaDosSantosSilva_DISSERT.pdf: 6981280 bytes, checksum: 0266d5d98ab38bb268fe5ddd9e7e3c97 (MD5) / Made available in DSpace on 2017-03-03T22:08:40Z (GMT). No. of bitstreams: 1 BrennaKarolynaDosSantosSilva_DISSERT.pdf: 6981280 bytes, checksum: 0266d5d98ab38bb268fe5ddd9e7e3c97 (MD5) Previous issue date: 2016-02-23 / A identifica??o de sistemas din?micos n?o lineares tem grande import?ncia em v?rias ?reas da Engenharia. Esta disserta??o apresenta uma t?cnica de identifica??o de sistemas utilizando Redes Neurais Artificiais do tipo Fun??es de Base Radial ? RBF aplicada a um sistema bastante comum na ind?stria de qu?mica e petroqu?mica, o sistema de multitanques (tanques acoplados), que apresenta uma din?mica de natureza n?o linear. Em tal sistema, o comportamento n?o linear deste pode dificultar a identifica??o e apresentar um modelo que n?o reflete seu comportamento real. Esse fator ? fonte de motiva??o para utilizarmos uma ferramenta matem?tica tal como a RBF na identifica??o dos aspectos n?o lineares do sistema. Essa rede apresenta potencialidades como habilidade de tratar sistemas complexos, aprendizado e generaliza??o. A identifica??o de sistemas utilizando RBF pode ser tratada como um problema de ajuste de curvas atrav?s da combina??o das fun??es de base. / The identification of nonlinear dynamical systems is of great importance in many areas of engineering. In this dissertation presents a system identification technique using Artificial Neural Networks Radial Basis Function type - RBF applied to a very common system in the chemical and petrochemical industry, multi tanks system (coupled tanks), which features a dynamic nature not linear. In such a system, the nonlinear behavior of this can make it difficult to identify and present a model that does not reflect their actual behavior. This factor is a source of motivation we use a mathematical tool such as RBF in the identification of nonlinear aspects of the system. This network has potential as ability to handle complex systems, learning and generalization. Identification systems using RBF can be treated as a curve fitting problem by combining the basic functions.