Identificação e controle de processos não lineares utilizando redes neurais artificiais

AUTOR(ES)

Adilson Jose de Assis

DATA DE PUBLICAÇÃO

2001

RESUMO

Considering that most of the industrial processes of interest to Chemical Engineering present a certain degree of inherent non-linearity or one introduced by systems of automatic control, an important necessity of inquiring about the performance the new techniques derived from artificial intelligence, whose interest here is in the artificial neural networks, capable of dealing with non-linearity in a straightforward way. It was made in this work a thorough bibliographical review related to the identification and control of non-linear systems. The various possibilities of identification of dynamic systems using parametric empirical models were presented according to a unifying view, with emphasis in the methods based on artificial neural networks. The main control techniques developed for non-linear processes as well as the main uses reported in literature on the Chemical Engineering field were also thoroughly reviewed. Afterwards, making use of two processes typical ofthe Chemical Engineering, (1) two tanks reactors connected in series, in which a exothermal reaction occur, with thermal exchange; (2) double effect evaporator; various possibilities of identification and control using neural networks were discussed, in several levels. The results, obtained through computer simulation, show the potential usage of neural networks (in the form NNARX and NNSSIF), especially in the techniques of predictive control, where the best results were obtained. The first system taken into account has complex dynamics and Single Input, Single Output (SISO), whereas the second system has Multiple Input, Multiple Output (MIMO). Uniting techniques derived from artificial intelligence, such as artificial neural networks, classic methods of identification and the modem theory of control, it was shown that these methodologies can be successfully used in the search for better performances of chemical processes under the action of automatic control

ASSUNTO(S)

redes neurais (computação) controle de processo - automação controle de processos quimicos reatores quimicos evaporadores identificação de sistemas

Documentos Relacionados

• Controle adaptativo de sistemas dinâmicos não-lineares utilizando realimentação de saída e redes neurais artificiais.
• Identificação remota de sistemas operacionais utilizando análise de processos aleatórios e redes neurais artificiais
• Identificação de cargas lineares e não-lineares em sistemas elétricos residenciais usando técnicas para seleção de atributos e redes neurais artificiais
• Controle de sistemas não lineares atraves de redes neurais
• Identificação de Madeiras utilizando a Espectrometria no Infravermelho Próximo e Redes Neurais Artificiais