作　　者： (邓娅莉）; (王吉芳）; (谷玉海）; (刘玉龙）;

机构地区： 北京信息科技大学现代测控技术教育部重点实验室

出　　处： 《北京信息科技大学学报（自然科学版）》 2017年第4期35-38,共4页

摘　　要： 为了更好地消除噪声成分,提取振动信号故障特征信息,利用双树复小波变换对采集信号进行分解,分别计算分解后各层信号的自功率谱及FFT幅值谱,并采用软阈值法对自功率谱进行筛选处理,对满足阈值条件的频率点所对应的FFT幅值谱中的频率点信号进行IFFT变换,变换结果经双树复小波变换重构获得去噪信号。实验结果表明,具有平移不变性和较好的保留信号细节信息的双树复小波变换,在结合自功率谱的阈值筛选方法消除随机噪声信号后,获得了较好的去噪效果。 To eliminate the noise components to extract the fault feature information in vibration signal,the sampled signal is decomposed by the dual-tree complex wavelet transform. Then the auto power spectrum and FFT amplitude spectrum are calculated with each decomposed layer signals.Further,special frequency points are selected from the auto power spectrum by the threshold approach.The special frequency points of FFT amplitude spectrum corresponding to the selected points from auto power spectrum are then transformed by IFFT. Finally,the IFFT calculation result is reconstructed by the dual-tree complex wavelet transform to generate the denoised signal. The experimental results indicate that the DTCWT has the translation invariance and can maintain the details of signals well;combined with the threshold selection method of auto power spectrum,it can eliminate random noise signals and obtain good effectiveness.

关 键 词： 双树复小波变换 降噪 阈值法 平移不变性