I'm trying to understand some data from a vibration sensor, in units of milli-g and in cartesian coordinates. One idea I had was to analyze the data in something like spherical coordinates, to understand if most of the vibration is radial, or how much is transverse along the angle dimensions. Can the FFT algorithm be used on data converted to spherical coordinates? Or are there similar algorithms for converting time series spherical coordinates to frequency domain? Can the FFT be used on the radial coordinate?

I have a physics background, but its been awhile. I'm trying to remember somethings from old courses and wanted to check if this is valid.

  • $\begingroup$ How many axis do you read? Do you only want a 1D FFT on the magnitude of your vibration? $\endgroup$ – Pier-Yves Lessard Feb 1 at 18:39
  • $\begingroup$ Maybe. I'm still trying to understand what the most important "dimension" is. I'm also hoping, if it is radius/magnitude, then the sensor gathering the data can be configured to only send that data-point, and reduce the data transmission needs. $\endgroup$ – Craig Skinfill Feb 1 at 20:34
  • $\begingroup$ I don't think there is a benefits of doing an fft on polar coordinate, you may be better to do a 2d fft (if you have 2 axis) and look at the magnitude plot. If you find a peak in the fourier domain, its location will have an angle relative to an axis, the will indicate the direction of the oscillation $\endgroup$ – Pier-Yves Lessard Feb 2 at 16:27
  • $\begingroup$ a single vibration sensor captures 2d information of time and frequency, i doubt it would be relevant to spacialize it in on a sphere. i thought you meant 3d modelling of sound spherically around a point, that would be logical. $\endgroup$ – com.prehensible Feb 2 at 16:37
  • $\begingroup$ The particular sensor (really just an accelerometer) captures acceleration data in X,Y, and Z axis. But gathering and transmitting that much data is slow. I'm looking to be able to monitor mechanical systems and determine when maintenance is needed. So I'm really curious IF we can determine that radial, or one of the angular dimensions, contains the most information and reduce the amount of data we send for analysis. At this point I'm just playing around with some ideas and asking the community. $\endgroup$ – Craig Skinfill Feb 4 at 15:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.