University of Ottawa Rolling-element Dataset – Vibration and Acoustic Faults under Constant Load and Speed conditions – Spectrograms (UORED-VAFCLS-P1)
Published: 24 May 2023| Version 2 | DOI: 10.17632/65d7pmfzvx.2
Spectrograms are created from the raw data by using the short-time Fourier transform (STFT), while using a Hanning window to convert the accelerometer and microphone values into 2D images. For each raw data file, 400 spectrogram images are created with a signal length of 512. A total of 24,000 spectrogram images are created, 12,000 images for each of the accelerometer and microphone data files. The processed data are provided in the zip files as portable network graphics files (.png).
Steps to reproduce
The Python code to convert the raw data to spectrograms is provided in the first link below. The second link provides the pre-print of the article related to this dataset.
University of Ottawa
Acoustic Imaging, Mechanical Vibration