Applycation And Reseatch Of Predictive Control Based On Neural Network In The Production Of PVC

After the second world war, petroleum chemical industry growth is very large. The rapid development of petrochemical industry promotes the growth of PVC effectively. Though they are affected by the economic crisis and oil price, the total development is enormous. In recent years, besides the promotion of objective factors, the development and application of various advanced control technology plays a huge role in the development of PVC production. PVC is produced in polymerization kettle. The temperature of plymerization kettle is the most important parameter among all the factors influencing the reaction process and the control precision of temperature is related to product quality directly. Polymerization kettle needs to heat up before the reaction process in order to meet the requirements of the reaction. The polymerization process is strong exothermic reaction and the temperature has the characteristics of time-varying, large lag, nonlinearity. It is difficult to meet the requirements of production process control with the conventional control method, so we hereby put forward the study of advanced control method striving to achieve better control effect.This paper relies on the background of polyvinyl chloride (PVC) production line in a chemical plant in Qingdao and proposes the predictive control method based on neural network by analyzing the characteristic of temperature variation and control in PVC polymerization process. Neural network has great learning performance, robustness and strong modeling ability. It can approximate nonlinear system with any degree of accuracy and has a strong ability to adapt to the nonlinear and time-varying. The predictive control with good tracking performance and strong anti-interference ability is widely used because it demands for Mathematical model little and is able to deal with pure lag process directly and shows strong robustness for the error of the model, etc. To realize the high performance control of PVC polymerization kettle temperature,we can adopt both advantages on control at the same time by combining the neural network control and predictive control when they are applied to nonlinear system. Therefore, we propose the predictive control method based on neural network. Experiments show that predictive control based on neural network has great adaptability, robustness and interference resistance. It can achieve high performance control.In order to show the characteristics of PVC polymerization process and control characteristics better, this paper uses the MATLAB simulation software for simulation verification and study the configuration implementation on DCS. DCS has the characteristics of reliability, adaptability, friendliness, coordination, etc and it can also realize the complex control algorithm. This provides the method for further application of the more advanced control scheme in industrial practice.