Yang D., Hu Z., Yang Y., He X., 2013, Weak Magnetic Memory Signal Denoising Based on Cascaded Singular Value Decomposition, Chemical Engineering Transactions, 33, 37-42.
Magnetic memory method is widely used in finding and locating stress concentration zone of ferromagnets, which is of great importance in early diagnosis of ferromagnetic structures and components. Magnetic memory sensing signal is weak and easy to be corrupted by noise and interference. In such situation, it is difficult to distinguish the characteristic of the magnetic sensing signal, and the stress concentration zone is not easy to be distinguished. Singularity value decomposition (SVD) is a nonlinear filter method useful for signal denoising and enhancement. But the singularity values are very sensitive to noise, and traditional SVD cannot process signals contaminated by heavy noise. As the denoising ability of single SVD system is limited, a novel method called cascaded SVD system (CSVD) is proposed in this paper. The noisy signal is processed by the first layer of SVD firstly. The output of the first layer of SVD is filtered by a second layer of SVD. And the output of the second layer SVD has a signal to noise ratio (SNR) gain over the first layer output. The principle of CSVD is proposed and applied to the enhancement of magnetic memory signal measured from gear tooth surface with a tiny defection. The characteristic of the signal for distinguishing the defect zone is obvious enhanced and the result has advantage over that getting from wavelet denoising method.
