Blind Identification of Mechanical Systems with Modulated Periodic Input Signals

Vibration signal analysis is the most common tool for the health monitoring of rotating machines. Nevertheless, the efficacy of vibration-based approaches is limited by the complex transfer functions (TFs) of the considered mechanical system. In this paper, we present a methodology aimed at accurately determining the impulse response of the gearbox transfer function leveraging prior system information. Subsequently, a transfer function removal preprocessing step is employed on the vibration signal to precisely identify the individual contribution of the system elements and achieve source separation. The effectiveness of our approaches is validated through examination utilizing synthesized datasets at varying levels of signal-to-noise ratio (SNR).