基于RBF网络直接广义预测精轧宽度控制

Direct Generalized Predictive Control Method Based on RBF for Finishing Strip Width Control

为了进一步提高热连轧带钢成材率和宽度精度,在精轧末机架后一般配置了测宽仪,利用机架间张力对宽度动态控制,达到调整带钢宽度目的。但由于精轧末机架中心线到测宽仪的距离较远,使得系统具有大滞后性及时变性等特点,很难建立精确的数学模型。采用对模型要求宽松的广义预测控制方法,用RBF网络直接逼近控制律,省略了Diophantine方程的求解、矩阵的求逆,缩短了在线计算的时间,适用于快速的轧钢过程。仿真结果表明这种方法对精轧宽度这一大时滞、不确定性系统有较好的控制效果。

To increase the rate of the qualified products and the precision of the width continuous hot strip mill, it usually sets width gauge behind the last finishing roll stand, and it can reach the purpose of adjusting the strip width by adjusting the tension of the stands. Because of the long way of the last finishing roll central line to width gauge, the system usually have a large time delay and time variation, and it is difficult to establish an accuracy model. By using the predictive control method and the RBF network to approximate the control rule directly, the online computing time is decreased, the solving of Diophantine equatiorl and inverse of matrix is avoided, and it is suitable for the fast rolling process. The simulation result shows the effectiveness of the proposed method.