提出了一种新的基于分数Fourier变换的自适应信号降噪方法,该方法的核心是利用不同信号在分数Fourier域具有不同的相关性,从而由信号本身得到自适应的滤波器。这种方法克服了传统时域或者频域滤波器需要确定窗函数的位置和形状的缺陷。通过与小波变换构造的滤波器进行比较,以数值仿真信号及实测滚动轴承振动信号降噪两个算例证明了方法的有效性。
A newly adaptive noise reduction method based on fractional Fourier transform is presented, the core of which is the different correlativity of different signals in the field of fractional Fourier transform. This method could present an adaptive filter from the measured signals themselves, and avoid the defect of traditional filter where one must determine the location and amplitude of window function in time or frequency domain. Based on the comparison of this method with the filter con- structed by wavelet transform, the feasibility and effectiveness of new method is verified through simulated and measured signals.