Method for diagnosing fault in aero-engine bearing based on stft-incepnext

Abstract

To effectively identify faults in an aero-engine bearing, ensure a high rate of fault identification, and achieve excellent accuracy.SOLUTION: The method includes: step 1 of collecting acceleration signals under various operating conditions and fault states of a bearing as sample data; step 2 of first determining the number of sampling points for each sensor, performing data combination by permuting and combining data collected by sensors located at different positions in the same time series, and then converting the fused vibration signals into time-frequency diagrams by using a short-time Fourier transform; step 3 of performing data preprocessing in which the converted time-frequency diagrams are normalized using mean values and standard deviations after dividing the data into a training set, a test set, and a validation set according to fault types; step 4 of constructing an IncepNext model and performing training and validation thereon; and step 5 of diagnosing aero-engine bearing faults using the trained IncepNext model and obtaining diagnostic results.SELECTED DRAWING: Figure 1

Images