Diagnostics of rolling bearings using artificial neural networks

Abstract

The vibration of bearings No. 60,206 with radial clearance of 0…0.45 mm is investigated. Analysis of the frequency response (FR) of vibration acceleration reveals information frequencies of 6 Hz, 15 Hz, 26 Hz, and 46 Hz, which coincide with the calculated values. The dependence of the vibration acceleration amplitude on the radial clearance for the specified frequencies is determined. The possibility of rolling bearing wear estimation by vibroacoustic diagnostic methods using artificial neural networks is proved. Diagnostic accuracy with the 4:4-3:1 MI network ranges from 75.6% for a radial clearance of 0.15 mm to 100% for a clearance of 0 mm.

Fault diagnosis of machines operating in variable conditions using artificial neural network not requiring training data from a faulty machine

“…A naturally occurring artificial intelligence model for diagnostic applications is that of classifier models. This is problematic, however, because when using neural networks as classifiers, learning data recorded for the machine to be diagnosed and for predicted faults are required [16][17][18][19][20][21][22][23][24]. There is work related to the reduction of damage data in learning vectors [25].…”