Simulation Of Neural Network Control In Vacuum Tower

There are many uncontrollable factors in the productive processes of the atmospheric and vacuum distillation unit. The most difficult points is how to control the complexity,the relativity and the uncertainty of the controlled objects. In order to control the system concisely and effectively, the traditional control method can not meet the demand, and the advanced control method can be used to achieve this. Aiming at various problems existing in the control system of vacuum tower, the main research content of this paper is selecting the vacuum tower top temperature and vacuum furnace exit temperature as the controlled objects and expanding simulation research with the control method of neural networks.Neural networks can approximate the complex nonlinear function. And it is suitable for solving time-varying problems, such as the lag of the reaction, the slow control and the large fluctuation of the temperature in vacuum tower.First of all, this paper briefly introduces the research background and the development status of neural network, the principle of Radial Basis Function neural network and its advantages and disadvantages, which lays a theoretical foundation for the simulation.Secondly, the control effect of the neural network control for the single input and single output control system is analyzed. The vacuum tower top temperature is selected feed forward and feedback control, and the vacuum furnace exit temperature is selected for cascade control. Then the simulation experiments are carried out on them. As the high quality requirements of the object in the control process, the strong correlation of the system and the problem of the time variation of the object, the traditional PID control module and the Radial Basis Function adaptive neural network control module are designed, and the simulation is carried out in the MATLAB software. The results of the traditional method and the new method are compared. Simulation results show that the RBF adaptive neural network control system has better control effect than the traditional control system, and it can also solve many problems in the control system.The internal relationship of the atmospheric and vacuum distillation unit is complex.And many parameters can not be used as control parameters and are easily affected by other parameters. In order to solve the problem of strong coupling between the vacuum tower top temperature and vacuum furnace exit temperature, the decoupling control effect of RBF neural network control for double input and double output control system is studied and analyzed. The simple decoupling module and the RBF module are designed,and the simulation is carried out in the MATLAB software. Research results show that RBF decoupling control can get better control effect.