131
CAT IV Part 1 Signal Processing slidebook
Autocorrelation
Correlation
Statistics
Processing
The autocorrelation function is used to reveal periodicities in
waveform data.
The FFT is also used to reveal periodicities, but
autocorrelation keeps the data in time waveform format
We will also see that autocorrelation is effective at revealing
signals with low duty cycles
The autocorrelation function can be considered a measure of
the similarity a signal has with a time shifted version of itself*
Autocorrelation can be calculated by performing the inverse
Fourier transform of the power spectrum.
* Noise and vibration analysis Brandt page 71
Copyright 2005-13 Mobius Institute All rights reserved
Autocorrelation
Correlation
Statistics
Processing
Copyright 2005-13 Mobius Institute All rights reserved
2014 Mobius Institute All rights reserved
www.mobiusinstitute.com
CAT IV Part 1 Signal Processing slidebook
132
Correlation
Statistics
Processing
Autocorrelation
The remaining waveform contains the correlated signals.
Uncorrelated (random) data is removed.
Synchronous and non-synchronous periodic data survives,
unlike time synchronous averaged data
It is therefore useful for rolling element bearing analysis and
other applications where the periodic patterns are difficult to see
in the waveform.
It is very useful for identifying faults at a low periodic rate
relative to analysis bandwidth*
For example, faults in the cage of a bearing
Signals with higher periodic rates and higher duty cycles will
be revealed via the FFT
* Autocorrelation as an analysis tool James C. Robinson
Copyright 2005-13 Mobius Institute All rights reserved
Example
Correlation
Statistics
Processing
Copyright 2005-13 Mobius Institute All rights reserved
2014 Mobius Institute All rights reserved
www.mobiusinstitute.com
133
CAT IV Part 1 Signal Processing slidebook
Example
Correlation
Statistics
Processing
Copyright 2005-13 Mobius Institute All rights reserved
Example
Correlation
Statistics
Processing
Copyright 2005-13 Mobius Institute All rights reserved
2014 Mobius Institute All rights reserved
www.mobiusinstitute.com
CAT IV Part 1 Signal Processing slidebook
Example
134
Correlation
Statistics
Processing
Copyright 2005-13 Mobius Institute All rights reserved
Example
Correlation
Statistics
Processing
Copyright 2005-13 Mobius Institute All rights reserved
2014 Mobius Institute All rights reserved
www.mobiusinstitute.com
135
CAT IV Part 1 Signal Processing slidebook
Example
Correlation
Statistics
Processing
Copyright 2005-13 Mobius Institute All rights reserved
Example
Correlation
Statistics
Processing
Copyright 2005-13 Mobius Institute All rights reserved
2014 Mobius Institute All rights reserved
www.mobiusinstitute.com
CAT IV Part 1 Signal Processing slidebook
Example
136
Correlation
Statistics
Processing
Copyright 2005-13 Mobius Institute All rights reserved
Example
Correlation
Statistics
Processing
Copyright 2005-13 Mobius Institute All rights reserved
2014 Mobius Institute All rights reserved
www.mobiusinstitute.com
137
CAT IV Part 1 Signal Processing slidebook
Correlation
Statistics
Processing
Example: One side of the roller
Copyright 2005-13 Mobius Institute All rights reserved
Example: The other side of the roller
Correlation
Statistics
Processing
Copyright 2005-13 Mobius Institute All rights reserved
2014 Mobius Institute All rights reserved
www.mobiusinstitute.com
CAT IV Part 1 Signal Processing slidebook
138
Correlation
Statistics
Processing
Example
Copyright 2005-13 Mobius Institute All rights reserved
That concludes this section
Copyright 2005-13 Mobius Institute All rights reserved
2014 Mobius Institute All rights reserved
www.mobiusinstitute.com
