Using a wavelet function as the filter function of Fourier self-deconvolution, a new method of resolving overlapped peaks, wavelet-Fourier self-deconvolution, is founded. The properties of different wavelet deconvolut...Using a wavelet function as the filter function of Fourier self-deconvolution, a new method of resolving overlapped peaks, wavelet-Fourier self-deconvolution, is founded. The properties of different wavelet deconvolution functions are studied. In addition, a cutoff value coefficient method of eliminating artificial peaks and wavelet method of removing shoulder peaks using the ratio of maximum peak to minimum peak is established. As a result, some problems in classical Fourier self-deconvolution are solved, such as the bad result of denoising, complicated processing, as well as usual appearance of artificial and shoulder peaks. Wavelet-Fourier self-deconvolution is applied to determination of multi-components in oscillographic chronopotentiometry. Experimental results show that the method has characteristics of simpler process and better effect of processing.展开更多
针对齿轮故障诊断中采集到的振动信号常伴有噪声干扰且故障特征难以提取的问题,以傅里叶-贝塞尔级数展开(Fourier-Bessel series expansion,FBSE)为基础,提出了一种将FBSE和基于能量的尺度空间经验小波变换(energy scale space empirica...针对齿轮故障诊断中采集到的振动信号常伴有噪声干扰且故障特征难以提取的问题,以傅里叶-贝塞尔级数展开(Fourier-Bessel series expansion,FBSE)为基础,提出了一种将FBSE和基于能量的尺度空间经验小波变换(energy scale space empirical wavelet transform,ESEWT)相结合的齿轮振动信号降噪方法,即FBSE-ESEWT。首先,将采集到的齿轮振动信号利用FBSE技术获得其频谱,以替代传统的傅里叶谱,接着凭借能量尺度空间划分法对获取的FBSE频谱进行自适应分割和筛选,以精确定位有效频带的边界点。随后通过构建小波滤波器组得到信号分量并进行重构,以减小噪声和冗余信息干扰;然后,为捕捉到更全面的特征信息将处理后的信号进行广义S变换得到时频图,输入2D卷积神经网络进行故障诊断验证算法可行性。通过对Simulink仿真信号和实际采集信号进行实验,结果表明,相对于原始经验小波变换(EWT)、经验模态分解(EMD)等方法,FBSE-ESEWT具有更好的降噪效果,信噪比提高了13.96 dB,诊断准确率高达98.03%。展开更多
文摘Using a wavelet function as the filter function of Fourier self-deconvolution, a new method of resolving overlapped peaks, wavelet-Fourier self-deconvolution, is founded. The properties of different wavelet deconvolution functions are studied. In addition, a cutoff value coefficient method of eliminating artificial peaks and wavelet method of removing shoulder peaks using the ratio of maximum peak to minimum peak is established. As a result, some problems in classical Fourier self-deconvolution are solved, such as the bad result of denoising, complicated processing, as well as usual appearance of artificial and shoulder peaks. Wavelet-Fourier self-deconvolution is applied to determination of multi-components in oscillographic chronopotentiometry. Experimental results show that the method has characteristics of simpler process and better effect of processing.
文摘针对齿轮故障诊断中采集到的振动信号常伴有噪声干扰且故障特征难以提取的问题,以傅里叶-贝塞尔级数展开(Fourier-Bessel series expansion,FBSE)为基础,提出了一种将FBSE和基于能量的尺度空间经验小波变换(energy scale space empirical wavelet transform,ESEWT)相结合的齿轮振动信号降噪方法,即FBSE-ESEWT。首先,将采集到的齿轮振动信号利用FBSE技术获得其频谱,以替代传统的傅里叶谱,接着凭借能量尺度空间划分法对获取的FBSE频谱进行自适应分割和筛选,以精确定位有效频带的边界点。随后通过构建小波滤波器组得到信号分量并进行重构,以减小噪声和冗余信息干扰;然后,为捕捉到更全面的特征信息将处理后的信号进行广义S变换得到时频图,输入2D卷积神经网络进行故障诊断验证算法可行性。通过对Simulink仿真信号和实际采集信号进行实验,结果表明,相对于原始经验小波变换(EWT)、经验模态分解(EMD)等方法,FBSE-ESEWT具有更好的降噪效果,信噪比提高了13.96 dB,诊断准确率高达98.03%。
