Avaliação e análise da aplicação de redes neurais e do algoritmo do enxame de partículas na modelagem e otimização do processo de polimerização.

Abstract

Polymers possess a great economic importance in current industry. In order to improve the development of news technologies based on new polymer applications, it is necessary to attain more economical and efficient levels on the study of polymers and of their production processes. It is also required to enlarge their capacities for the development of novel polymers, which shall present the specific characteristics demanded by industry. The present work introduces a model for the development of new polymers, at low costs and with a higher flexibility level. This model was developed by means of designing, upgrading and training a neural net, employing the algorithm named “particle swarm” as objective function. A simulator model was established upon this neural net, whose utilization allows to identify the specific parameters which are to be considered as adequate to the optimization of reaction conditions in the process of polymer production.