【摘要】 以频域多参考点模态参数识别方法(frequency domain poly-reference,简称FDPR)为研究对象,通过矩阵的灵敏度分析及Kronecker代数给出该方法中模态参数方差误差的估计方法,提高了模态参数识别的可靠性。估计过程主要分为5步:a.通过相干函数估计频响函数中随机噪声的协方差;b.通过频响函数中随机噪声的协方差计算缩减频响函数的协方差;c.利用缩减频响函数的协方差估计系统矩阵的协方差;d.由系统矩阵的协方差估计极点的协方差;e.由极点的协方差估计模态参数的方差。最后,通过一个4自由度仿真算例对所给出的估计方法进行了验证。更多还原

【Abstract】 Variance error estimation for frequency domain poly-reference algorithm is illustrated based on matrix sensitivity analysis and kronecker algebra.The presented method included five steps:a.Estimating noise covariance in frequency response function through coherence function.b.Calculating noise covariance in reduced frequency response function.c.Estimating covariance of system matrix by noise covariance of reduced frequency response function.d.Estimating poles covariance through covariance of system matrix. e.The modal parameters variance is obtained by poles covariance.Finally,a simulated case of a four DOF system is used to demonstrate the accuracy and practicability of the presented estimation method.更多还原