The paper discusses a promising non-destructive method for detecting and classifying bearing defects using microwave sensors. The sensors irradiate a vibrating object with microwave waves, receive reflected waves, and extract phase measurements between them that correspond to the surface movement during vibration. We use the received signal to monitor the defects of rolling bearings with different parameters by the presence of characteristic high-frequency pulses in it. Since these pulses are usually noisy with extraneous masking vibrations, we use special signal processing. To do this, we obtain the envelope spectrum and examine it for signals with an increased excess value by scanning the spectral excess. Then, from the received signal, we determine the parameters of the bandpass filter and use it to select the desired signal. Based on the central frequency of this defective signal, we determine the defect itself after comparing it with the reference values.