小波多分辨率分析时变系统参数识别算法的鲁棒性研究

针对多自由度时变系统参数识别问题,基于Daubechies小波多分辨率展开的时变参数辨识方法分析影响参数识别鲁棒性的各个因素。通过数值分析针对突变、线性慢变以及谐波快变的时变参数进行识别,研究结果表明:当基函数dbN一定时,在预先确立的分解尺度范围内,识别精度随分解尺度的增加而增加;待识别参数的频率特性对分解尺度的选择有很大影响,快时变参数比慢时变参数对分解尺度更为敏感;基函数dbN并不是影响识别精度的主要因素;在分解尺度相同的情况下,可以通过提高采样频率增加快时变参数识别精度。

Robustness of a Parametric Identification Algorithm for Time-varying Systems Based on Wavelet Multi-resolution Analysis

Research,based on Daubechies wavelet multi-resolution analysis,was carried out to solve parameter identification problems in multiple degrees of freedom time-varying systems.In order to improve identification efficiency and accuracy,numerical experiments,based on the above method,were conducted to study the various factors that affect performance.The results show that when the basic function dbN was fixed in the preset decomposition scale,identification accuracy increased with an increase in the decomposition scale.The frequency component of the time-varying parameters had great influence on the choice of decomposition scale,and the fast time-varying parameters were more sensitive than the slow.The choice of the basic function dbN affects the identification accuracy,but is not a key factor; an increase in sampling rate can improve the identification accuracy of fast time-varying parameters under the same decomposition scale.