Styrene polymerization reaction is one of the most important chemical manufacture processes, but until now, the control for its product quality cannot be solved effectively. So, it is very important to model the reaction process firstly.Through the research on the mechanisms of styrene polymerization reaction, it has been known that the production processes are dynamic and products are of the distribution features. Therefore, a DRNN (Diagonal Recurrent Neural Network), which has the dynamic feature and a B-spline neural network with the distribution feature, are combined in the paper to model a polystyrene process. Meanwhile, out of the consideration of actual approximation, the white noises as disturbances are added into training data to set up models. Taking advantage of those simulating data coming from an experimental rig, the affectivity of the modeling method via combined neural networks is approved for the styrene polymerization processes.Genetic Algorithm (GA) is a kind of optimizing method,...
|