Model-predictive-control based on subspace identification and its application
基于子空间辨识的模型预测控制策略及其应用

To deal with the nonlinearity and time-varying characteristics in the processes of chemical industry, an adaptive-predictive-control strategy based on the recursive subspace identification is proposed. The predictive models obtained from the subspace identification are considered the initial models, which are compared with the online updated model to generate matching errors. The model with the smallest matching error is selected for use in calculating the process control input, thus improving the model accuracy. The control simulations of a simulated moving bed(SMB) show that the method is robust to the system parameters perturbation and efficient in attenuating external disturbance.