Cyclostationary Beamforming for the Acoustic Localization of the Failed Bearing

The significant cyclostationary characteristic of vibration signal resulting from bearing pitting failure is widely applied in vibration signal-based fault diagnosis. However, specific scenarios pose challenging issues regarding the proper installation of vibration sensors. It is worth noting that the cyclostationarity of the vibration signal causes a corresponding cyclostationarity of the corresponding acoustic signals. Consequently, non-contact fault diagnosis of rolling bearings can be achieved through acoustic-based diagnosis technology. Sound field visualization is a common approach that combines micro-phone array measurements with acoustic beamforming. Locating the failed bearing is possible by visualizing the cyclostationary characteristics of the region of interest. This paper introduces the cyclostationary beamforming as an innovative way to localize failed bearings in a transmission system. Specifically, the cyclic spectral density (CSD) is used to reflect the cyclostationarity of the region of interest at a specific cyclic frequency. By obtaining steering vectors and cross cyclic spectral matrix (CCSM) and focusing on the CSD of each scanning point, we can localize the failed bearing accurately. Numerical simulations and a gearbox experiment have been conducted to demonstrate the effectiveness of the proposed method for localizing the failed bearing.