CONCENTRIC FUZZY ENTROPY FOR THE FAULT DIAGNOSIS OF CYLINDRICAL ROLLER BEARING

Effective fault diagnosis is essential for ensuring operational safety and preventing catastrophic failures in rotating machinery. Recently, entropy-based techniques have gained prominence as valuable tools for feature extraction. Among these, fuzzy entropy has attracted significant attention due to its robustness in processing nonlinear signals. Nevertheless, fuzzy entropy encounters limitations in accurately identifying the fault severity of roller bearings. Fault characteristics generally exhibit a broadband spectral distribution, with critical discriminative information often located in sidebands near characteristic frequencies. The Haar wavelet’s restricted spectral coverage limits its capacity to capture diverse structural oscillations within fuzzy entropy, leading to suboptimal feature extraction. To overcome this limitation, this study introduces a novel feature extraction approach termed concentric fuzzy entropy. This method utilizes a multi-wavelet strategy to achieve comprehensive fault feature extraction across the entire frequency spectrum. Employing this methodology, a diagnostic framework for cylindrical roller bearings attains a fault severity identification accuracy of 93 percent, outperforming five other entropy-based benchmarks in experimental evaluations.