Soft-Sensing Modeling Method of Vinyl Acetate Polymerization Rate

In the procedure of polyvinyl alcohol production, it is hardly accurate measuring the Vinyl Acetate (VAC) polymerization rate. To solve this awkward situation, this paper adopt Genetic Algorithm-Back Propagation (GA-BP) Neural Network to fit the nonlinear relation of the variables which derives from the production process by setting the VAC polymerization rate as the master variable and the initiator addition ratio, methanol ratio, polymerization temperature and VAC activity degree as auxiliary variables. Establish VAC polymerization rate soft-sensing model based on GA-BP network which the connection weights optimized by genetic algorithm. The comparison results with BP network based on the actual measured data show that the model this paper constructed is accurate and effective.