在研究谐波小波频段分解与Hilbert解调分析的基础上，提出了基于谐波小波包变换的解调分析法，并详述了其具体实现过程。该方法根据齿轮箱频谱特征与齿轮箱故障特征频率的理论计算值，确定所需提取的特征啮合分量，并在此基础上选择谐波小波包分解层数与所需提取的频带带宽；通过傅里叶变换及其反变换提取出相应的特征啮合分量，然后借助Hilbert算子对提取出的啮合分量进行包络解调分析。采用这种方法对存在磨损及点蚀故障的齿轮箱振动信号进行了分析，结果表明，基于谐波小波包的包络解调法具有精确提取任意频段调制信息的能力，能够为齿轮箱故障源的准确定位及故障程度提供可靠的判断依据。

On the basis of partial decomposition in frequency domain with harmonic wavelet package method and Hilbert demodulation， the demodulation based on harmonic wavelet package transform is introduced. The Fourier spectra analysis is carried out after preprocessing， and the depth of decomposition and frequency band are determined by the analysis of frequency band distribution and the theoretic computation of gear box characteristic frequency. Then with FFT and IFFT the corresponding meshing component is extracted. Finally， the modulation information is extracted via Hilbert envelop analysis approach. With the proposed method， wear and notch faulty gearbox are tested and analyzed by experiment， respectively. The results prove that this method can locate gearbox fault precisely and provide corresponding basis for judging fault cause and degree.

国家自然科学基金资助项目(51005261)；重庆市科委自然科学基金计划资助项目(CSTS，2009BB0355和 2009BB319)