基于正交投影的闭环子空间辨识方法

A Closed-Loop Subspace Identification Method Based on Orthogonal Projection

针对闭环MIMO系统的辨识问题,提出一种基于投影的子空间辨识方法,可以辨识出具有白噪声干扰的控制对象和控制器的状态空间模型。该方法从控制对象的输入输出来辨识对象模型。通过正交投影使得求得的状态实质为卡尔曼滤波状态,以此来求取控制对象的状态空间模型;对于控制器模型,提出两种不同的辨识求解方法。最后,在Matlab下进行了仿真验证,结果表明该方法的有效性。

A new subspace identification method for MIMO closed-loop system is proposed, which can identify the state space model of a plant with white noise and controller. Based on input and output data, the method can get system state through orthogonal projection, and the state is essentially Kalman filter state, and then calculate state space model of the plant; to the controller model, the paper present two different identification approaches. At last, the simulation results show the good effectiveness.