Research On Decoupling Control On Temperature Of PVC Stripping Tower Based On Neural Network

Polyvinyl chloride (PVC) industry occupies an important place in national economy. Vinyl chloride monomer mixed in PVC after polymerization is an important factor affecting the quality of PVC, and vinyl chloride is toxic. So the stripping process after polymerization is necessary. Improving the temperature control accuracy of PVC stripping process has a great significance on improving the quality of PVC products, saving energy, reducing production and reducing environmental pollution.PVC stripping process is a nonlinear, coupled multi-variable system with large time delay, and using the traditional methods will lead to unstable product quality, low control accuracy and low capacity. So the deep research on control of stripping process is made in this paper.Firstly, we analyze the characteristics and control requirements of PVC stripping process. Based on these, we use dynamic fuzzy neural network establishing the model of stripping tower by actual operating data of stripping process of Jinhua group. Dynamic fuzzy neural network is characteristic of universal approximation and ability to learn, requiring less prior knowledge, and can learning online. The simulation results show that the model is valid.Secondly, stripping process is a coupled nonlinear system through analyzing the model of stripping process in detail. A neural network decoupler is designed based on that it does not need the model of control object to be accurate. Stripping process is turned into two single-variable system from a dual-input dual-output coupling system by using neural network doupler. The simulation results prove that the neural network decoupler is effect.Finally, a neural network PID controller is designed for the single-variable system of stripping process after decoupling. The controller adds a link of neural network control to adjust the parameters of PID controller compared to traditional PID controller. At last the neural network PID controller is connected to the neural network decoupler for the decoupling control of the stripping process. The simulation results show that the neural network decoupler proposed in this paper has a good decoupling and robustness.The simulation results show that this decoupling control system proposed in this paper not only has a simple structure, easy to implement, but also has a strong decoupling and robustness. This method provides a new way on improving the control precision of the stripping process.