连续时间模型的闭环子空间辨识

Closed-loop Subspace Identification for Continuous-time Models

本文对直接使用采样数据进行连续系统的闭环子空间辨识问题进行了研究.将线性滤波方法与基于主元分析的子空间辨识相结合,利用参考输入或者外部激励信号的高阶滤波变换的正交投影变量作为辅助变量,提出了一种新的连续时间系统闭环子空间辨识算法.数值仿真表明了与其他算法相比,本文提出的算法具有很好的辨识效果.

The problem of closed-loop subspace identification for continuous-time models from sampled data directly is considered. Combining subspace identification based on principal component analysis and linear filter method, a novel closed-loop subspace identification algorithm for continuous-time model is proposed by using the orthogonal projection variable of high order filter transform of reference inputs or exogenous inputs. The identification performance of the proposed algorithms is illustrated by numerical simulation comparing with other algorithms.