Singular Spectrum Decomposition Based Nonlinear Energy Operator for Rolling Bearing Faulty Diagnosis

Abstract

Fault signals of rolling bearings are non-stationary and the fault characteristics are often submerged in the background noise. Unfortunately, the fault characteristic frequency detection from composite fault signals of rolling bearings is difficult. To address this problem, during this paper, an improved method is proposed, which combines the singular spectrum decomposition (SSD) with the nonlinear energy operator (NEO). SSD can separate a hybrid signal into several sub-signals in accordance with their different frequencies adaptively and greatly alleviate mode mixing, etc. The decomposed components name singular spectrum components (SSCs) whose optimal component comprises fault features. Meanwhile, NEO is an excellent tool to demodulate modulating signals with enhancing amplitudes advantage. Subsequently, the selected SSC is demodulated by NEO. Toward upon testifying the effectiveness of the enhanced method, comparison investigation with EMD is also given in the paper. The comparison result validly testifies the improved method is more effective to detect fault characteristic frequency.