针对常规控制方法在模型比较复杂的化工生产过程中控制效果不佳且过程具有非线性时变的特点,提出一种新型的滚动窗口自适应子空间预测控制方法.由输入输出数据通过子空间辨识方法获得子空间预估器,将子空间预估器作为预测模型,设计一种滚动窗口自适应预测控制器,并引入决定系数来过滤不良数据.通过对2-CSTR系统过程控制的仿真试验,验证了所提出控制方法的有效性.
In order to deal with the problems that the control performance is not satisfied when the conventional control methods are applied to the chemical processes and the nonlinear,time-varying characteristics in the processes,a novel receding window adaptive subspace predictive control method has been proposed.The subspace predictor,obtained through the subspace identification algorithm from input-output data,is regarded as the prediction model to design the receding window adaptive predictive controller.Meanwhile,the coefficient of determination has been introduced to filter the bad data.The process control simulations of a 2-CSTR system show the effectiveness of the proposed method.