短时傅里叶变换的时频聚集性度量准则研究

发布时间：2018-04-05 10:19

本文选题：短时傅里叶变换　切入点：时频分析　出处：《振动.测试与诊断》2017年05期

【摘要】：为了使短时傅里叶变换(short time Fourier transform,简称STFT)获得良好的时频聚集性,必须根据信号的具体特点来选择窗长。分析了现有窗长选择准则的选择机理及其优缺点,发现归一化3阶Renyi熵准则与Stankovic准则一般只会取到窗长选择范围的最大值,并分析了出现这种问题的原因,而其他准则得到的也不是最优的窗长。提出了一种新的基于对数窗能量的窗长选择准则。对数窗能量与窗长是一种非线性关系,这显著区别于普通窗能量随窗长线性增长的特性,且其增长速度与窗型无关,并在短窗和长窗具有不同的增长速度,因而能够在短窗和长窗之间取得良好的折衷。提供了仿真信号和实际信号的处理实例,其结果证明对数窗能量准则使STFT获得了良好的时频聚集性,效果优于现有的窗长选择准则。
[Abstract]:In order to obtain good time-frequency aggregation of short time Fourier transform (STFT), the window length must be selected according to the characteristics of the signal.The selection mechanism, advantages and disadvantages of the existing window length selection criteria are analyzed. It is found that the normalized third-order Renyi entropy criterion and the Stankovic criterion can only obtain the maximum window length selection range, and the reasons for this problem are analyzed.Other criteria are not optimal window length.A new window length selection criterion based on logarithmic window energy is proposed.Logarithmic window energy is a nonlinear relation with window length, which is obviously different from the characteristic that the ordinary window energy increases linearly with window length, and its growth rate is independent of window type, and has different growth rates in short window and long window.Thus, a good compromise can be achieved between short and long windows.Examples of simulation and actual signal processing are provided. The results show that the logarithmic window energy criterion makes STFT obtain good time-frequency aggregation, and the effect is better than the existing window length selection criterion.
【作者单位】：华南理工大学机械与汽车工程学院;
【基金】：国家自然科学基金资助项目(51275174,51375178) 广东省自然科学基金资助项目(S2012010008789)
【分类号】：TN911.7

【相似文献】