Bearing run-to-failure datasets of UNSW

Published: 4 November 2021| Version 3 | DOI: 10.17632/h4df4mgrfb.3
Siyuan Zhuang,
Zhongxiao Peng


The run-to-failure experiments data were collected at the University of New South Wales in 2019-2020, regarding the development of bearing fault severity assessment methods. The following journal paper [1] contains the description of the experiments and test rig information. [1] H. Zhang, P. Borghesani, R. B. Randall, and Z. Peng, “A benchmark of measurement approaches to track the natural evolution of spall severity in rolling element bearings,” Mech. Syst. Signal Process., vol. 166, p. 108466, 2022, doi: 10.1016/j.ymssp.2021.108466. Please consider citing this resource if relative to your work. Data of four run-to-failure experiments, named Test 1 to 4 according to Ref. [1], are stored in four folders respectively. Each folder contains two subfolders, one has all collected signals at 6 Hz, and the other has selected samples at 4 speeds (6,12,15,20 Hz). Each data file is named as vib_aaaaaaaaa_bb.mat: • aaaaaaaaa: is the running cycle of the shaft in this measurement • bb: is the shaft speed in Hz Each data file contains the following variables: • Fs: the sampling frequency • accH: acceleration (V) in horizontal direction • accV: acceleration (V) in vertical direction • enc1: encoder signal channel 1 (1024 pulses/rev) • enc2: encoder signal channel 2 (1024 pulses/rev) • loadCell: the applied radial load (V) on the test bearing • tacho: tacho signal of the shaft (1/rev) For any further information, please contact the Author or the Tribology and Machine Condition Monitoring group of UNSW Sydney. The version of datasets is updated due to the nomenclature issue of variables.



University of New South Wales


Mechanical Vibration, Vibration Condition Monitoring