作　　者： ; ; ; ;

机构地区： 燕山大学

出　　处： 《中国机械工程》 2007年第22期2659-2662,共4页

摘　　要： 在冷带轧机电液伺服系统中应用广义预测控制理论直接算法,采用随机梯度估计法在线估计控制器参数模型,避免了在线求解Diophantine方程,减少了在线计算时间。该控制算法基于离散受控自回归积分滑动平均模型,吸取了自适应控制的优点,使控制系统具有较好的鲁棒性和智能性。仿真研究结果表明,该方法比目前带钢生产中广为采用的PID控制策略优越,使得系统的动态品质有了较大的改善,响应时间和控制精度都有所提高。 The direct algorithm of generalized predictive control was applied in the electro--hy- draulic servo system of cold rolling mill, which adopted stochastic gradient estimate algorithm to estimate parameters of the controller on--line. It avoided on--line solving the Diophantine equation and reduced the on--line calculation time. Based on the discrete controlled autoregressive integrated moving average（CARIMA） model, the control algorithm was combined with merits of the adaptive control, which made the control system possess the excellent robustness and the intelligent capability. The simulation result indicates that its control effect is better than that of PID control which is comprehensively used in strip production. The adopted control strategy makes the dynamic quality of the system be improved largely. The response time and the control precision are improved too.

关 键 词： 广义预测控制 冷带轧机 电液伺服系统 受控自回归积分滑动平均模型 随机梯度估计法

领　　域： [金属学及工艺] [自动化与计算机技术] [自动化与计算机技术]