Prior Knowledge Enhanced-Velocity Guided Chirplet Transform for Nonstationary Feature Estimation with Application to Bearing Fault Diagnosis in Vehicle Electric Drive Systems

Effective diagnosis of electric bearing faults, especially weak early-stage faults, remains challenging due to complex rotational-speed conditions and external noise. Although the velocity-guided chirplet transform (VGCT) can effectively characterize nonstationary components, it requires synchronized rotational-speed information, which is often difficult to obtain in practice. To this end, this paper proposes a prior-knowledge-enhanced velocity-guided chirplet transform (PKVGCT) for fault diagnosis without prior rotational-speed information. PKVGCT includes two stages. In the first stage, multiple candidate instantaneous frequencies (IFs) are obtained via peak searching and are smoothed by a second-order difference operator to support obtaining a reliable guideline. In the second stage, the manually selected guideline is further used in the analysis of VGCT, and simultaneously a weak-noise energy filtering mechanism is applied to ultimately yield a high-quality time-frequency (TF) representation (TFR). In numerical analysis, the qualitative analysis of the TFR results, the mean relative error (MRE)-based quantitative analysis, and the robustness analysis under different noise levels all demonstrate the superior performance of PKVGCT. With two experimental cases, the effectiveness of PKVGCT in practical applications is fully demonstrated.