AGC-ASC Decoupled Neural Networks Predictive Control Method

The coupling models for the thickness-crown objects is established. A Dynamic Matrix Controller based on the TH neural networks is given with the conver-gence property. The computer simulations with the AGC-ASC decoupled neural networks predictive control system is complemented and it shows that the stable states of neural networks are reached with on more that one μs, this has not only satisfied the fast prop-erty of rolling process, but also obtained a higher control index.