How to measure vibration frequency with the output (Voltage) from accelerometer?

SH Yang
SH Yang Registered Users Posts: 9
Hi,
Actually, I need to measure the actual wind turbine vibration after attaching a couple of accelerometers on a wind turbine and a spare blade.
So.. I have been trying to understad the processes for a week.
However, I still do not know how the voltage outputs from accelerometer will be used for vibration analysis (vibration frequency.)?
I will use piezoelectric accelerometers.
Below process is what I have found.
Accelerometer data -> Sampling Rate & Max Bandwidth -> FFT -> Data Analysis -> Output Diagostic Algorithm.

Comments

  • jagec
    jagec Solar Expert Posts: 157 ✭✭
    Re: How to measure vibration frequency with the output (Voltage) from accelerometer?
    SH Yang wrote: »
    Hi,
    Actually, I need to measure the actual wind turbine vibration after attaching a couple of accelerometers on a wind turbine and a spare blade.
    So.. I have been trying to understad the processes for a week.
    However, I still do not know how the voltage outputs from accelerometer will be used for vibration analysis (vibration frequency.)?
    I will use piezoelectric accelerometers.
    Below process is what I have found.
    Accelerometer data -> Sampling Rate & Max Bandwidth -> FFT -> Data Analysis -> Output Diagostic Algorithm.

    The piezo device will output a variable voltage as it flexes.

    Back in the old ways, you would use this voltage to position the pen on a plotter, and you would get stacks and stacks of paper with squiggly lines on it. By increasing the speed of the paper, you would be "sampling" at a higher rate, which would allow you to capture higher-frequency vibrations. But that was limited by the mechanics of the pen.

    These days we measure the voltage directly, and digitize it using an analog-to-digital converter. The sampling rate (how many measurements of voltage you take per second) determines how high of a frequency you can observe...in other words, if your turbines are vibrating back and forth 12 times a second, and you're only taking a measurement once a second, you can't accurately determine the vibration frequency. So, you have to sample at a higher rate than the highest frequency you expect to encounter. Sampling theory says that if you sample at TWICE the rate of the highest frequency, you don't lose any data. In other words, if we want to capture that 12 times a second vibration, we would take a measurement 24 times a second (we won't get into compressed sensing).

    Now you have a file that consists of a bunch of voltage and time measurements. To convert this to frequency, we do a Fast Fourier Transform (FFT)--this is a computer algorithm that takes our time-domain data, and puts it into the frequency domain. Think about looking at AC power on an oscilloscope. We see a bunch of up-and-down curves that corresponds to the voltage fluctuations. If you Fourier transform that data, you will see a SINGLE peak at 60 Hz, since that is the frequency at which AC (in the US, at least) oscillates. Now if we take the same AC output, but we add a resistor that gets connected and disconnected once a second, the Fourier transform of that oscilloscope reading would have TWO peaks, one at 60 Hz and the other at 1 Hz.

    That is overly simplified, but hopefully it helps. Further analysis of the Fourier data can involve thresholding (if the signal isn't above a certain level, ignore it), band-pass filtering (if the signal is above a certain frequency, or below a different frequency, ignore it), and some other functions. But just looking at the raw, unprocessed Fourier transform will probably be enough to tell you what you need to know about wind turbine oscillations--after all, if the oscillations aren't big, they probably aren't going to wreck the turbine.
  • BB.
    BB. Super Moderators, Administrators Posts: 33,433 admin
    Re: How to measure vibration frequency with the output (Voltage) from accelerometer?

    I would search around for information on:

    Helicopter
    Forced Response Vibration Analysis


    Similar mechanics and responses for a Wind Turbine (blades, hub, gear boxes, relatively smooth source of power (gas turbine) vs load (typically three phase alternator).

    Are you looking at static analysis (nothing moving/rotating) or under operation?

    -Bill
    Near San Francisco California: 3.5kWatt Grid Tied Solar power system+small backup genset