In time-frequency filter design based on S transform, the time-frequency filtering area can be selected by the Exp, and the basic wavelet of S transform is fixed. To solve these problems, the concept of information entropy theory was introduced into the generalized S transform time-frequency filtering, and an adaptive generalized S transform time-frequency filter design method based on the spectral entropy was proposed. While calculating the signal' s time-frequency spectrum by generalized S transform, the time-frequency spectrum' s frequency range at each time point can be divided into some small zones; and while calculating the interval spectral entropy, the mean value of time-frequency spectrum in the spectral entropy maxima interval can be taken as the threshold to measure signal' s disorder degree. The spectrum value at this time point can be cleared while it' s greater than threshold. The de-noised signals can be reconstructed by generalized S inverse transform. This method can be adaptive to adjust time-frequency filtering factor and effectively remove interference information. Applying this method and a conventional filtering method to simulating the signal processing can verify the validity of this method.%将信息熵引入到广义S变换时频滤波中,提出一种基于谱熵的自适应广义S变换时频滤波器设计方法.计算信号广义S变换的时频谱,将时频谱各时间点的频率区间划分成若干个小区间,计算各区间时频谱熵,以时频谱熵最大区间的时频谱的平均值作为阈值,以此来衡量信号的杂乱程度.若该时间点上的时频谱值大于阈值,则将该点时频谱值清零.采用广义S逆变换进行重构.该方法可自适应地调节时频滤波因子,有效滤除干扰信息.
展开▼
