Analysis of initial conditions for polymerization reaction using fuzzy neural network and genetic algorithm

In order to determine initial conditions for preparation of polybutadiene with given physicochemical characteristics, a fuzzy neural network (FNN) model was constructed to estimate the physicochemical characteristics of the polymer (the ratio of cis form polymer and the polydispersity index (PDI) and the conversion ratio from the initial conditions in the batch polymerization process. The mean absolute errors of the FNN model for the conversion ratio, the ratio of cis form polymer and PDI as the actual scale were 7.13, 0.23 and 0.17%, respectively. Analyzing for the constructed FNN model, the relationships between the process conditions and physicochemical characteristics were obtained as IF-THEN rules. Using the constructed FNN model and a genetic algorithm (GA) combined with reliability index (RI), the process conditions with the given physicochemical characteristics and conversion ratio were calculated. The calculated and actual process conditions showed an average relative error of 3.9%.