Controle em destilação batelada : controle adaptativo e controle preditivo com modelo baseado em redes neurais artificiais

AUTOR(ES)

Ana Maria Frattini

DATA DE PUBLICAÇÃO

1995

RESUMO

Separation processes are very important in chemical industries where batch distillation appears as one of its oldest processes. Over the last few years batch processing has received renewed attention because of the growing interest in the production of high-technology, high-value added chemical and biochemical products and also in environmental problems of industrial waste minimization. Based on similar phenomenological aspects, both batch and continuous distillation are characterized as non-linear processes, but due to the absence of feed streams in batch equipments the temperature and composition profiles become instantaneous, with substantially different control systems requirements. In order to control top composition, by handling the reflux ratio, digital control systems taking into account the unsteady process characteristics were developed where design did not follow the conventional feedback control loop by measuring bottom composition or column pressure differential, therefore minimizing the energy consumption. An ordinary way of adjusting the controller parameters was initially studied: the Programmable Adaptive Controller (PAC). The adjustments are previously scheduled and on-line performed through measurements of an auxiliary variable which correlates well with the changes in process properties and on the involved process phenomena. In practice, however, where complex mixtures are usually processed, the independence on the referred knowledge about the mixture is crucial and claims an on-line process identification procedure... Note: The complete abstract is available with the full electronic digital thesis or dissertations

ASSUNTO(S)

destilação inteligencia artificial controle de processo

Documentos Relacionados

• Controle adaptativo de uma coluna piloto de destilação em batelada com inferenciação de composição atraves de redes neurais artificiais
• Simulação e controle preditivo linear (com modelo de convolução) e não-linear (com modelo baseado em redes neurais artificiais) de colunas recheadas de absorção com reação quimica
• FCC : controle preditivo e identificação via redes neurais
• Suporte à decisÃo para vigilÃncia epidemiolÃgica baseado em modelo preditivo de surtos de dengue utilizando redes neurais artificiais
• Controle adaptativo de sistemas dinâmicos não-lineares utilizando realimentação de saída e redes neurais artificiais.