Intelligent Decoupling Control Of Cold Rolling Thickness And Tension System

Thickness and shape are the two major quality indicators in the modern coldrolling. AGC (automatic thickness control) system must have a high accuracy pressureregulation system and control system of the support in order to obtain high accuracyof product thickness. Tension is an important process parameters in strip rolling, eachrack built various links by the tension. It needs to study the variation of tension andthe relationship between various factors mathematical model to properly control therolling process. In the strip rolling process, it exists more serious dynamic couplingbetween amount to be adjusted and adjustment amount because of the fluctuation oftension or plate thickness, preventing the independent control of each variable effect.Sometimes even undermine the system work, so they can not put into operation. As aresult, the decoupling control was introduced into rolling mill systems.In this paper, five-stand tandem cold mill in the background, on the thickness andtension multivariable system were emphasized, the main research contents and resultsare as follows:Firstly, we analyzed the features of nonlinearity, strong coupling, uncertainty andmulti-constraint of the thickness and tension system and built a dynamic engineeringmathematical model of thickness and tension system. Which based on the existingfield data obtained thickness and tension system transfer function, and use MATLABsoftware to be discrete, as the application of decoupling algorithm behind thefoundation.Secondly, the advantages and disadvantages of several commonly used algorithmof decoupling control are studied and analyzed. It verifies the reasonableness of thistopic and necessity through MATLAB verified system does exist the strong coupling.Finally, the established mathematical model for the conventional PID control,research and simulation results show that its decoupling control is better, but theanti-interference ability is poor. Therefore, it is proposed a multivariable PID controlalgorithm based on quasi-diagonal recurrent neural network (QDRNN), and itsapplication to the thickness and tension multivariable decoupling control system.Decoupling control results show that the composite algorithm is decoupled responsespeed and anti-jamming capability significantly better than the traditional PIDdecoupling control.