Quantitative diagnosis of rolling bearings based on dual-impulse extraction of acoustic emission signals

Rolling bearings are used utilized in various applications and have serious damage consequences, so fault diagnosis is very important. When the rolling element passes through the spall area, it will generate two impulses at the time of entry and exit. Extracting these impulses to calculate the spall size can help identify the degree of damage, achieve quantitative diagnosis, and provide a guidance for maintenance and replacement. Acoustic emission technology is widely used in bearing state detection and fault diagnosis due to its unique sensitivity to early faults. It has greater potential in quantitative diagnosis. However, it is susceptible to noise which may interfere with the observation of early dual-impulse phenomenon to some extent. Thus, this paper has proposed solutions for above features: modified cepstrum to remove electromagnetic noise while retaining the fault impulses, spectral averaging negentropy to avoid the influence of random impulses and select the optimal demodulation frequency band. Simulation and test signals validate the advantages of this method, and the expansion of spall area is observed through bearing degradation tests.