中文摘要：

机械故障检测过程中，由于反映机械故障的振动信号微弱，很容易被外界噪声干扰信号污染，从而影响机械故障诊断。为提取纯净振动信号，传统EEMD滤波算法虽具有较强的降噪能力，但由于EEMD算法存在缺乏严谨理论基础、运算效率低、容易造成有用信号丢失等缺点，致使降噪效果不理想。为解决以上问题，提出一种基于变模式分解和频谱特性的自适应降噪算法。基于变模式分解优点，通过分析有用信号模态与噪声模态频谱特性，提取有用信号模态从而实现降噪。通过仿真信号与实测信号分析表明，新算法降噪效果优于传统EEMD滤波算法。

英文摘要：

In the process of mechanical fault detection,the vibration signal reflecting the fault feature is weak and is easyto contaminate by outside noise,which increases the difficulty of diagnosing the mechanical fault. In order to extract the pure vi?bration signal and solve the above problem,an adaptive denoising algorithm based on variable mode decomposition and spec?trum characteristics is proposed,because the traditional ensemble empirical mode decomposition(EEMD) filtering algorithmlacks rigorous theoretical foundation,has low operation efficiency and is easy to make the useful signal lost,which may cause apoor denoising effect although it has strong ability of denoising. On the basis of the advantages of variable mode decomposition,the useful signal mode is extracted to achieve denoising by means of analyzing the spectrum characteristics of useful signal modeand noise mode. The analysis results of the simulation signal and the measured signal show that the new algorithm is superior tothe tradition EEMD in denoising.