A fuzzy algorithm for flatness control in hot strip mill

Based on BP neural network, a flatness prediction model in hot strip mill was developed, in which the same location point data were adopted for training and testing to avoid the influence of time-delay. Two fuzzy flatness control algorithms in hot strip mill were developed, and the simulations were carried out by using the flatness prediction model as controlled objective. Conventional fuzzy control algorithm replacing PID linear control system entirely can reduce flatness error significantly. However, the control quality is not good in steady state. The coupled fuzzy-PID control algorithm reduces flatness error significantly as well as produces the desired flatness with small steady-state error and good stability. The coupled fuzzy-PID control algorithm is found to be suitable for rolling the desired flatness in hot strip mill.