Fault detection via sparsity-based blind filtering on experimental vibration signals

Detection of bearing faults is a challenging task since the impulsive pattern often fades into noise. Moreover, tracking health conditions rotating machinery generally requires characteristic frequencies components interest, which can be cumbersome constraint for large industrial applications because extensive number machine components. One recent method proposed in literature addresses these difficulties by aiming to increase sparsity envelope spectrum vibration signal via blind filtering (Peeters. et al., 2020). As name indicates, this no prior knowledge about machine. Sparsity measures like Hoyer index, l1/l2 norm, and spectral negentropy are optimized approach using Generalized Rayleigh quotient iteration. Even though has demonstrated promising performance, it only been applied signals an academic experimental test rig. This paper focuses on real-world performance sparsity-based complex challenges ensure numerical stability convergence optimization. Enhancements thus made identifying quasi-optimal filter parameter range within tackles issues. Finally, certain frequency ranges order prevent optimization from getting skewed dominant deterministic healthy content. The outcome proves that filters effective without any frequencies.