提出了一种基于正交投影的滚动窗口递推闭环子空间的辨识算法.它引入正交投影(QR)分解来进行正交投影、特征值分解,从而实现奇异值分解,并使用了bona fide递推算法来更新QR分解.然后根据投影变量选取原则选用了两个不同的投影变量W1和W2,推导并提出了基于正交投影的两种递推闭环子空间辨识方法.最后,数值仿真表明:提出的递推闭环子空间辨识方法是有效的、可行的. An orthogonal projection based scroll window recursive closed-loop subspace identification algorithm is proposed, which introduces orthogonal projection (QR) decomposition for orthogonal projection and eigenvalue decomposition to achieve singular value decomposition, and uses bona fide Recursive algorithm to update QR decomposition.According to the selection principle of projection variables, two different projection variables W1 and W2 are selected and two kinds of recursive closed-loop subspace identification methods based on orthogonal projection are deduced and proposed.Finally, numerical simulation shows The proposed recursive closed-loop subspace identification method is effective and feasible.