基于HWPT-ZFFT的二维全息谱计算方法
Research of 2-D Holospectrum Calculation Based on HWPT-ZFFT Method

精确的频率、相位和幅值识别是进行全息谱计算的必备条件，针对含3个及以上的密集频谱成分，提出一种基于谐波小波包变换的频谱细化方法（harmonic wavelet packets transform-zoom fast Fourier transform,简称HWPT-ZFFT），较传统的复调制细化傅里叶变换所利用的低通及带通滤波器相比，其盒型频谱特性可将感兴趣频段的信号正交，冗余、泄漏地提取出，提高了识别精度。首先，利用谐波小波包对密集频谱成分进行滤波；然后，频移进而重采样，进行傅里叶变换得到细化的频率、幅值及相位；最后，计算密集频率下二维全息谱，进行双盘转子全息谱计算，考虑高次分倍频，得到更丰富的故障特征。仿真及双盘转子实验结果表明所提出方法的有效性。
A HWPT-ZFFT (harmonic wavelet packet transform and zoom fast Fourier transform) method is proposed for spectrum zoom, targeted for a spectrum that contains three or more intense frequency components. When the signals are orthogonally decomposed into different frequency bands with no redundancy or leakage, they perform better using the box-like spectrum instead of the tradition low-pass and band-pass filter in complex modulation ZFFT. First, the HWPT method is employed to filter the intense components, shift the center frequency to zero, and resample. Then, the frequency, amplitude and phase after zoom are obtained by FFT in order to calculate the holospectrum. Finally, rich fault characteristics are obtained under higher frequency. The effectiveness of this method is demonstrated through simulation and multi-span rotor system experimental results.