MNAD

Minimum Noise Amplitude Deconvolution

MNAD is a blind deconvolution algorithm for enhancing repetitive fault impulses, and similar algorithms include CYCBD, MCKD, MOMEDA, etc. We defined the periodic noise amplitude ratio (PNAR) metric as the minimization criterion for MNAD, and solved MNAD by the backward automatic differential blind deconvolution (BADBD) algorithm we proposed earlier. MNAD can adaptively locate one or more resonance bands excited by local faults. The paper also compares the performance of MNAD and the optimal resonance band selection algorithms (Autogram, IESFOgram).

We use five experimental data to verify the performance of MNAD. Content includes:

MNAD: The implementation function of the MNAD algorithm;

resultPlot: The function for plotting time-domain waveform, spectrum, and envelope spectrum;

Demo_MNAD: main function;

105.mat; signal_Seu.mat; 59.csv; signal_IMS.mat: test signal.

We also provide a tensorflow based implementation of MNAD (https://github.com/FangBo-0219/MNAD.git) which is significantly faster to compute.

Reference

[1] B.Fang, et al., Minimum noise amplitude deconvolution and its application in repetitive impact detection, Structural Health Monitoring,2022,Accepted.

[2] B. Fang, J. Hu, C. Yang, Y. Cao, M. Jia, A blind deconvolution algorithm based on backward automatic differentiation and its application to rolling bearing fault diagnosis, Meas. Sci. Technol. 33 (2022) 025009. https://doi.org/10.1088/1361-6501/ac3fc7