ABSTRACT
Metal Magnetic Memory (MMM) technique is a fast way to locate the stress concentration zone of the ferromagnetic material, which can diagnose mal-function in the earlier stage. The MMM signal is weak, and can be easily effected by the various factors such as environment interference and electronic noise, as such it becomes difficult to exactly locate the stress concentration zone. Considering the stationary of the MMM signal, two kinds of signal feature extraction methods are proposed. Firstly, the combination of the MMM signal peak-peak, gradient is proposed in time domain, which can help identify the stress concentration zone and be easily used in engineering applications. Then, according to the non-stationary and singularity of the MMM signal, the MMM signal energy after decomposition has been enhanced based on Teager energy operator of wavelet coefficients, the multi-scale related feature is extracted for the low signal-to-noise ratio signals, that accurately determines the stress concentration. Finally, the proposed method is proved to be effective through the experimental data.
