一种改进的特征系统实现算法及在智能结构中的应用

An Improved Eigensystem Realization Algorithm and Its Application to Modal Parameter Identification of Intelligent Space Trusses

针对多输入／多输出特征系统实现算法（ Ｅ Ｒ Ａ），提出了一种改进算法，这种算法通过以递推形式形成一个对称半正定阵，采用特征值分解代替奇异值分解，并采用相关滤波的方法对测量数据进行预处理以提高识别精度。本文还将其用于具有密集模态的空间智能桁架的模态参数识别中，以建立受控数学模型，对改进算法运算量的理论分析与应用实例表明，该算法的运算速度可提高５～１０ 倍。

An improved Eigensystem Realization Algorithm,which can increase computational efficiency greatly and is called Fast ERA(FERA), is proposed in the paper. In the approach,a symmetric positive semidefinite matrix is formed recursively and the singular value decomposition is replaced by eigenvalue decomposition.Furthermore,the test data are preprocessed by correlation filtering to improve the accuracy of modal parameter identification.The algorithm is used to identify modal parameters of a space intelligent truss with closely spaced frequencies for constructing a mathematical model for vibration control.Theoretical analysis and simulations demonstrate that the algorithm can synthesize an identical model to that found by the Eigensystem Realization Algorthm (ERA),but at the computational speed increased by 5～10 times for large data set sizes.The improved algorithm is a better Multi input & Multi output identification method.