Broadening the application of angle-time cyclostationary analysis in bearing fault diagnosis using precise speed reconstruction

Angle-time cyclostationary (AT-CS) analysis has developed into the most powerful analytical tool for studying vibration signals of rotating machinery under variable operating conditions. However, due to limitations in the accuracy of traditional rotational speed estimation algorithms, the effectiveness of AT-CS analysis may be compromised in the absence of robust keyphasor signal. To address this challenge, a novel rotational speed signal estimation method based on “double estimation” strategy is proposed to expand the application scenarios of AT-CS. This methodology entails an initial estimation of the rough speed through time-frequency analysis, followed by refinement using phase demodulation techniques. Additionally, the Vold–Kalman order tracking method is employed to isolate the single frequency signal component essential for phase demodulation techniques. In situations where the spectrum is complex and the speed fluctuates violently, this method can still accurately obtain the speed signal, achieving similar effects to the rotational velocity transducer. The complex speed curve experiment and turbine pump bearing experiment with severe speed fluctuation are carried out to verify the algorithm.