Hot-rolling Board With The Thickness Of The Neural Network Prediction Model And Fuzzy Control Methods Research

Thickness precision is one of the most important quality indexes of strips. Automatic Gauge Control System (AGC) is an important part in automatic system in mill. It is difficult to build an exact mathematic model because of the complexity of mill process and the uncertain parameters. Intelligent technology is an effect way to solve the problem. The main contents of this thesis are as follows:Based on the spring equation, the P-h graph was used to analyse the errors of lengthways strip thickness. The principle of automatic gauge control was analysed.A thickness prediction model in hot strip mill was developed based on BP neural networks. The Trial-and-Error method was used to confirm the best number of hidden neurons. Compared with the models that with or without traditional mathematical model of spring equation as input, a hybrid model which combined BP neural network with mathematical model was developed. Compared with the data collected from factory fieldwork, the relative error was below 1%. High-precision prediction of rolling thickness was achieved.The mathematical model of electric-hydraulic position control system was obtained by using the method of mechanism modeling. Two kinds of controllers were designed and applied to control the hydraulic AGC system, which are PID controller and fuzzy self-tuning PID controller. The performance of these two controllers was compared based on simulation experiments. The simulation results show that the capability of fuzzy PID controller is much better than traditional PID controller. The system is more robust base on the Fuzzy-PID control theory.