冷轧机液压AGC系统GPC控制仿真研究

Simulation Research on Generalized Predictive Control of Hydraulic AGC System of Cold Strip Mills

冷轧机液压AGC系统已经成为现代带钢轧机中控制带钢厚度的关键设备,其动态性能具有大惯性、大时滞、非线性等特点。通过参考典型的液压元件的作用机理,建立了冷轧机液压系统的数学模型。利用基于支持向量机(SVM)的广义预测控制(GPC)算法对液压系统进行控制,构成了液压AGC计算机仿真系统,仿真结果表明:SVM学习速度快,在小样本情况下具有良好的非线性建模和泛化能力;基于SVM的GPC算法具有很好的控制性能。

Hydraulic automatic gauge control （AGC） system has become one of key equipments to control exit strip thickness in strip rolling mills, which has dynamic performance of big inertia, time lag and nonlinear so on. Mathematics model of hydraulic system is founded in whole based on the action principle of some representative hydraulic components. Hydraulic system is controlled with algo- rithm of Generalized Predictive Control （GPC） based on Support Vector Machine （SVM）, which forms simulation system of hydraulic AGC by computer and whose simulation results show that SVM can be trained quickly and has good ability of modeling nonlinear process and good generalization under small data set available; GPC algorithm based on SVM model shows satisfactory performance.