Application of an Improved Minimum Noise Amplitude Deconvolution for Bearing Fault Diagnosis

• Published in: IEEE access Vol. 12; pp. 174182 - 174192
• Main Authors: Ma, Ying, He, Siming, Qi, Entie, Wei, Yu
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 2024; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Accuracy; Amplitudes; Autocorrelation; autocorrelation analysis; Bearing fault diagnosis; Deconvolution; Effectiveness; Fault detection; Fault diagnosis; Filtering algorithms; Finite impulse response filters; improved MNAD; improved sparrow search algorithm; minimum noise amplitude deconvolution; Noise; Parameter estimation; Parameter robustness; Robustness; Search algorithms; Vibrations
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2024.3501583

The detection of cyclic impulsive components, which serve as crucial indicators for extracting bearing faults, from vibration signals holds significant importance in fault diagnosis. The deconvolution methods have been demonstrated as a useful tool for highlighting cyclic impulsive components induced by bearing faults. Recently, a novel deconvolution technique named Minimum Noise Amplitude Deconvolution (MNAD) was proposed to effectively enhance the periodic impulses from heavily corrupted signals. The challenges, however, exist in the application of the MNAD under harsh working conditions. The challenges primarily arise from the rigorous requirements for the multi-input parameters. To address these issues, an improved MNAD (IMNAD) is proposed in this study. First, the novel approach uses the autocorrelation analysis of the envelope signal to estimate a key parameter of the fault period, instead of depending on the given prior period. Moreover, an improved sparrow search algorithm combining sine-cosine and Cauchy mutation (SCC-SSA) is employed to determine the optimal values for the remaining two key parameters, namely the filter length, the number of iterations, and the noise ratio. The IMNAD, in comparison to the original MNAD, exhibits adaptability in parameter selection across diverse operational conditions, thereby demonstrating its efficacy and robustness. Finally, the effectiveness and superiority of IMNAD are validated through simulated and two real bearing fault signals.