关键词: 模态参数识别, 极大似然估计, 参数不确定性, 连续时间模型, 正交多项式

Abstract: A multiple input/output frequency domain modal parameters identification method, which takes stochastic noise into account, is proposed. This method belongs to maximum likelihood class estimation series, and is based on the rational fraction model of continuous time domain frequency response function. Forsythe orthogonal polynomial is utilized to improve the numerical condition. First, the least square estimation is implemented to get the initial value of modal parameters. Then, with the "a priori" stochastic noise information, the optimization process of maximum likelihood estimation is carried out to get better identification result. Moreover, uncertainty information of the modal parameters is obtained, according to the Cramer-Rao low bounder inequality, almost without additional calculations. The GARTEUR aircraft model is taken to validate this algorithm.

Key words: modal parameters identification, maximum likelihood estimation, parameter uncertainty, continuous time domain model, orthogonal polynomial