摘要
闭环辨识算法具有广泛的工程应用前景,而子空间方法近年来应用于多个领域中,但子空间方法无法直接应用于闭环辨识,因此研究闭环子空间辨识算法具有重要意义。两步方法可用于辨识闭环系统,但计算量巨大,推导复杂,需要进一步改进。针对这种情况,提出了一种改进算法,使用将两步方法与正交投影相结合的方法,并利用QR分解实现,直接构建虚拟信号序列,大大减少了计算量,最后使用PI-MOESP辨识算法辨识模型。仿真实验将该算法与其他子空间辨识算法相比较,显示出该算法的有效性及计算量的显著减少。
Closed- loop identification is widely applied in engineering applications, but subspace identification method can＇ t be used for closed - loop identification directly, so the research on closed - loop subspace identification method is very important. Two -stage method can be used for closed -loop identification, but this method has large amounts of computation. Because of this, an improved method is presented. This method combines two - stage method with orthogonal projection, and uses QR decomposition for numerical implementation, uses PI - MOESP algorithm to identify model at last. Simulation shows that this method is effective.
出处
《计算机仿真》
CSCD
2008年第7期302-305,共4页
Computer Simulation
关键词
闭环辨识
子空间方法
两步方法
正交投影
Closed-loop identification
Subspace method
Two-stage method
Orthogonal projection
