A new denoising method based on dual-tree complex wavelet transform and nonlinear time series was proposed,considering the weakness,such as the phase distortion,of the wavelet soft-threshold denoising method,in which the real and image parts of the coefficient are processed individually.The new method process the magnitude of the complex coefficients instead,taking into account the fact that the magnitude does not oscillate in positive and negative directions which is more suitable for threshold denoising and the fact that the coefficients of the fault signal are always periodic.The nonlinear time series method can be used to strengthen the periodicity of the coefficients caused by the fault signal and to restrain the noise meanwhile.In the method proposed,the fault signal was decomposed by dual-tree complex wavelet transform to obtain the coefficients of different layers,the nonlinear time series method was used to strengthen the periodicity of the coefficient,and then the soft-threshold denoising was carried out to remove the DC component.Finally, the fault characteristic signal was obtained by coefficient reconstruction.The simulation and experimental results show the effectiveness of the method,and a new efficient denoising method was provided.%针对双树复小波变换传统软阈值降噪方法对实、虚部树系数分别进行阈值处理时提取的强背景噪声下轴承故障特征信号效果不理想，且实、虚部分离的阈值处理方法会引起局部相位失真问题，利用故障信号小波变换系数具有周期性与双树复小波系数模震荡小等特点，提出双树复小波变换与非线性时间序列方法相结合的强背景噪声下轴承故障特征提取方法。对故障信号进行双树复小波变换，获得各层小波系数并求模，选择系数模周期性较强层系数进行非线性时间序列处理，增强系数中周期性成分，抑制随机噪声；对增强后系数进行软阈值处理消除直流成分对提取结果的影响；将处理后系数还原为复数形式进行双树复小波重构，可成功提取弱故障特征信号。仿真、试验信号处理结果表明，该方法能有效提取强背景噪声下的故障特征信号。
展开▼
