论文部分内容阅读
提出一套适用于噪声环境的飞机颤振模态参数辨识方法。为减小噪声对辨识结果的影响,首先设计了一种针对扫频激励的时频滤波器,利用扫频信号及其响应在时频域分布较为集中的特点,有效去除噪声,提高了试验数据的信噪比。为进一步提高辨识精度,提出了一种基于随机模型的频域广义最小二乘辨识算法。将噪声条件下的系统辨识问题转化为广义整体最小二乘问题,并采用线性的广义奇异值分解求解模型系数,避免了非线性优化的复杂计算。通过优化加权项,获得了接近极大似然估计的辨识效果。最后,通过试飞试验数据验证了方法的有效性。
A set of flutter modal parameters identification method suitable for noise environment is proposed. In order to reduce the influence of noise on the recognition result, a time-frequency filter for swept-frequency excitation is designed. By using the characteristics that the sweep signal and its response are more concentrated in the time-frequency domain, the noise is effectively removed and the test data is improved The signal to noise ratio. In order to further improve the recognition accuracy, a generalized least squares identification algorithm based on stochastic model is proposed. The problem of system identification under noisy conditions is transformed into the generalized least squares problem, and the linearized generalized singular value decomposition is used to solve the model coefficients, which avoids the complicated calculation of nonlinear optimization. By optimizing the weighted terms, the identification effect of near-maximum likelihood estimation is obtained. Finally, the flight test data verify the effectiveness of the method.