Enhanced condition monitoring of variable-speed bearings via adaptive TQWT-based signal purification

To address the signal distortion caused by environmental interference in bearing fault detection under time-varying operational parameters, this study develops a multi-level adaptive thresholding approach integrating adjustable quality-factor wavelet decomposition (TQWT) with envelope spectral diagnostics. The proposed framework implements three-stage signal processing:(1) Performing multi-resolution decomposition on acquired vibration data through TQWT to extract multiscale components; (2) Designing nonlinear threshold operators using sigmoidal functions for optimized high-frequency coefficient processing; (3) Executing signal reconstruction through synthesized fusion of modified high-frequency elements and retained low-frequency constituents. Experimental validation conducted on a variable-condition test rig demonstrates the dual capability of this methodology in suppressing stochastic noise while preserving critical fault signatures. The quantifiable results confirm the method’s efficacy under variable operating environments, substantially improving diagnostic reliability for industrial rotating machinery.