A Fuzzy Neutral Network Controller Based on Optimized Genetic Algorithm for UC Rolling Mill

Considering the mathematical model for the control system of intermediate bending roll in UC rolling mill is time-varying and uncertain, the conventional PID algorithm can´t achieve a rapid and accurate response when some parameters change, so that the precision of the sheet shape can´t be easily ensured. To realize the accurate control for the intermediate bending roll, a fuzzy neural network controller is designed and applied in the loop control system of the intermediate bending roll. The genetic algorithm is used for the optimization search in the process of network training, and then the BP network is adopted to get a high precision solution. The simulation result shows that compares to conventional BP algorithm in the network training, the fuzzy neural network controller proposes in this paper can achieve rapid response, small ultra regulation and a better robustness against model uncertainty.