一种改进的随机子空间辨识方法的稳定图
A Stabilization Diagram Based On the Alternative Stochastic Subspace Identification
为在模态参数辨识中更好地区分虚假模态与物理模态,对传统时域子空间方法的稳定图进行了改进。改进的稳定图利用了由输出信号的自相关函数重构的Hankel矩阵,认为系统阶次固定,模态特征的变化趋势随数据量的增加而体现。在稳定图中引入了可表征各阶估计模态贡献量大小的分量能量指标(CEI)作为判断模态特征稳定性的判据之一,以剔除虚假特征。通过对一个受噪声污染、具有密集模态的振动仿真系统进行辨识,可从改进的稳定图中可以看到物理模态随数据量增加仍保持稳定,且对应于较大的CEI值,虚假模态的表现则明显相反,验证了该改进的子空间方法的稳定图的正确性和有效性。
To separate spurious and physical modes, the classical stochastic subspace identification method had been alternated. The traditional Hankel matrix was replaced by a reformed one. Accordingly, the alternative stabilization diagram shows the variation or stabilization of the modal parameters with the row increments of the reformed Hankel matrix. The so-called component energy index (CEI), which indicates the vibration intensity of signal components, had been used as a pole stability criterion. In the stabilization diagram, spurious modes are of low CEI and unstable, while the physical ones show opposite performance. More spurious modes could be removed and the diagram illustrates the stable poles more clearly. It helps to attain more accurate estimates. The proposed stabilization diagram is performed to a simulated 4-DOF system with two close modes and small signal to noise ratio (SNR). The estimated results demonstrate that the alternative one has helped to provide an effectual and practical tool for output only modal parameters identification.
模态参数辨识 / 随机子空间方法 / 稳定图 / 分量能量指标CEI {{custom_keyword}} /
modal parameters identification / stochastic subspace identification / stabilization diagram / CEI {{custom_keyword}} /
/
〈 | 〉 |