Adaptive residual spectral amplitude modulation: A new approach for bearing diagnosis under complex interference environments

Rolling bearings are a vital component for transmitting and supporting in rotating machinery. They are susceptible to failure since their operation under high speeds and heavy loads. Bearing failure may disrupt the manufacturing process and cause catastrophic accidents. Therefore, condition monitoring for bearings is essential to minimize operational disruptions and avoid unforeseen casualties. High-frequency resonance demodulation technology (HFRDT) and many improved approaches provide an effective rolling bearing's fault diagnosis tool. Rotating machinery tends to be sophisticated in Industry 4.0, while workplace interference escalates. As a result, fault information and interference components are coupled within the same frequency band range, rendering linear filter HFRDT-based methods ineffective. Therefore, spectral amplitude modulation (SAM) is proposed to nonlinearly enhance the fault information for bearing diagnosis. However, interference components dominate the monitor signal in complex environments, making it challenging for SAM to achieve a satisfactory result in practical applications. Therefore, an adaptive residual spectral amplitude modulation (ARSAM) approach is proposed for diagnosing bearing under complex interference environments. In this work, the raw signal is separated into multiple narrowband signals within various frequency bands. Then, residual spectral amplitude modulation (RSAM) is performed on the narrowband signal to obtain fault information nonlinearly. Subsequently, a different fault harmonic-to-noise rate (DFHNR) index is presented to select the optimal signal with rich fault information adaptively. Lastly, the simulated signal and real engineering data are analyzed to showcase the performance of the ARSAM. The results indicate the superior performance of the proposed approach in recognizing bearing faults under complex working environments.