Intelligent Decoupling And Control Research Of Looper Height And Tension In Hot Continuous Rolling Mill

Looper system is a key part in the process of the hot strip, its control effect is very important for the product quality and the stability of the production process. This paper established the mathematical model of looper system, and then put forward two control schemes for the decoupling and control of system. Finally, the simulation experiments were completed in the MATLAB simulation environment, and the results show that the two control schemes have good dynamic and static response and strong anti-interference ability. Paper specific content as follows:(1) According to the actual operation parameters of a hot rolling factory,the approximate linear coupling model of looper system was established in the vicinity of system working point.This paper analysed the overall stability of the looper system with linear system theory and the coupling degree of system with relative gain method,and then pointed out the relative gain method shortcomings and the coupling characteristics of looper system.(2) Two fractional order PID controllers controlled the looper system on the basis of decoupling with the method of the single neuron decoupling,and the control parameters of system were fixed online.(3)The auto-disturbances-rejection technology was introduced to the decoupling and control of the looper system.The system was controlled and decoupled dynamically by two parallel auto-disturbances-rejection controllers on the basis of static decoupling, and the key parameters of system were fixed online.(4)A kind of improved genetic algorithm with a new mutation operator based on the preselection mechanism was put forward in this paper.Nine complicated functions were introdeced to test the performance of the improved genetic algorithm,and the test results showed that the improved algorithm has a high convergence speed and a good global search ability.The improved genetic algorithm combined with the gradient descending method to be a new algorithm.The new algorithm was used to train two RBF neural networks offline that were used to identify the jacobian information of looper system online,so two networks could get a good initial condition for training online.Considering that the controllers had many parameters,this paper used the improved genetic algorithm to optimize control parameters, and then got better initial values of control parameters, so that greatly reduced the difficulty of setting parameters.