基于改进经验包络法的非线性系统参数识别

Parametric identification of nonlinear system based on improved empirical envelope method

基于调幅调频分解的经验包络法广泛用于线性和非线性系统参数识别。然而,在调幅调频分解中,为获得纯调幅调频信号而进行的多次包络迭代,会增大累积误差。在经验包络法中,求导信号不光滑而引起的过包络,会增大系统参数识别误差。分别采用滑窗阈值去噪思想与滑动平均技术解决上述两个问题,提出了改进的经验包络法,并基于该方法对单自由度非线性系统自由衰减振动模态参数识别。通过几个算例参数识别分析,证实了该方法具有良好的抗噪声性能与较高的识别精度。

The empirical envelope method based on amplitude modulation-frequency modulation(AM-FM)decomposition is widely used in parametric identification of linear and nonlinear systems.However,in AM-FM decomposition,multiple envelope iterations to obtain pure AM-FM signal can increase cumulative error.In the empirical envelope method,the over envelope caused by unsmooth derivative signal can increase error of system parametric identification.Here,the sliding window threshold denoising idea and the moving average technique were used to solve the above two problems,respectively to propose an improved empirical envelope method.Based on this method,free damped vibration modal parameters of a single-DOF nonlinear system were identified.Through parametric identification analyses of several examples,it was shown that the proposed method has good anti-noise performance and higher identification accuracy.