Polymeric reactor is the main unit of fluoroelastomer product line, in which the reactor temperature is the most important controlled variable that will affect the product quality dramatically. However, it is hard to achieve the satisfactory reactor temperature control performance by using the traditional control method. Because polymeric reactor is a non-linear, large delay, large inertia plant and the polymeric chemical reaction of the flouring-containing monomers is too complex. On the basis of the powerful logical reasoning and expression ability of fuzzy theory, the ability of on-line self-learning neural network, the adaptability of complex industry process of prediction control have been applied in the studying of fluoroelastomer production process and polymeric chemical reaction of the flouring-containing monomers.In the first the thesis introduces the characteristics and controlling strategies of the polymerization reaction, then analyzes the characteristics of control object and the problems of conventional control programs. According to the uncertainty and non-linear characteristics of control object, the idea of using QNN to replace the traditional PID control is proposed. According to the characteristics of intermittent operation, a method of using Section-switching control is put forward. At different stages of polymerization reactions, according to their characteristics, corresponding control programs is applied. The results showed that the QNN controller is superior to the original PID program, its stability and the temperature overshoot are in line with technological requirements.This paper design and develop the polymeric reactor temperature Quantum Neural Network predictive control system and use the control system to produce fluoroelastomer. The polymeric reactor temperature is illustrated to be controlled in desired range, then the product quality is improved. |