论文部分内容阅读
奇异值分解(SVD)是信号处理的一种有效方法,通过构造Hankel矩阵,依据奇异值差分谱和奇异值的逼近度量原则,构造出的最佳逼近矩阵求逆得到去噪后的新的信号序列。通过与二次样条小波,墨西哥草帽小波,多贝西小波进行对比,基于SVD方法能有效检测出信号的奇异性,克服小波变换的相位偏移。通过比较SVD和小波分析方法对列车振动信号分解重构,SVD重构能有效滤除噪声。通过功率谱比较,SVD重构信号能真实反映0-5Hz频段的振动特性。
Singular value decomposition (SVD) is an effective method of signal processing. By constructing the Hankel matrix, based on the principle of approximation of the singular value difference spectrum and the singular value approximation, the optimal approximation matrix is constructed and the new denoised signal is obtained sequence. By comparing with quadratic spline wavelet, Mexican straw hat wavelet and Doppelian wavelet, the singularity of signal can be effectively detected based on SVD method and the phase shift of wavelet transform can be overcome. By comparing SVD and wavelet analysis to decompose train vibration signals, SVD reconstruction can effectively filter out noise. By comparing the power spectrum, SVD reconstruction signal can truly reflect the vibration characteristics of 0-5Hz band.