Cepstrum Lifting and Coherent Averaging-Based Instantaneous Frequency Estimation of Low-Amplitude Signal Components for Rotating Machinery

Instantaneous frequency (IF) serves multiple critical functions in the prognostic and health management of rotating machinery. The current IF estimation methodologies primarily concentrate on enhancing estimation accuracy at the algorithmic level. However, in practical industrial applications, the target signal component is often influenced by varying operating conditions, structural resonances, and background noise, causing it to become submerged or even disappear, ultimately resulting in the failure of conventional IF estimation techniques. To address these challenges, a novel IF estimation approach incorporating signal quality enhancement is developed. First, the complementary multisource vibration signal is processed by the proposed segmented cepstrum lifting (CL) method to overcome the submersion of the resonant frequency band on the target component. Then, the processed segmented signals are fused using coherent averaging (CA) technology to enhance the amplitude of the target component. Finally, leveraging the first-order approximation capability of the short-time fractional Fourier transform (STFrFT) for nonlinear frequency modulation modes, precise IF estimation is achieved. Verified by field data, the proposed method outperforms mainstream methods in both accuracy and applicability, highlighting its industrial application prospects.