辨识与控制分离的广义预测控制算法

ADAPTIVE GENERALIZED PREDICTIVE CONTROLLER SEPARATING IDENTIFICATION FROM CONTROL

提出了一种自适应广义预测控制策略．其中，对象参数辨识及预测模型参数计算等以较大的采样周期进行，滚动优化控制以较小的采样周期进行．将辨识与控制运算分别放在２个不同的中断服务程序中，使辨识与控制分离，通过不同采样周期的模型变换建立起辨识与控制的联系，可大大缩短控制运算时间。

In this paper, a new adaptive generalized predictive control strategy is proposed; plant parameter estimation and predictive model parameter recursion computation and so on are executed in a long sampling period, and roll optimization control calculation is carried out in a short sampling period. In addition, identification and control calculation are put in different interrupted routine separately. It can shorten control calculation time, and therefore, a shorter control period can be adopted.