FFT Analysis of Vibration Information

lcdi

FFT Analysis of Vibration Information [Copy link]

 

From the perspective of signal processing, since the amplitude fluctuation is large and difficult to stabilize, this factor will be considered later. (Amplitude variation law)
Detecting the frequency domain information of vibration is more likely to achieve the goal.

Specifically, the function option ARMA in the expert mode is adaptive filtering
. The other functions reflect amplitude information.
Only FFT can obtain frequency domain information. So, carefully study the FFT function of the sensor.

The following figure shows fast jitter

The following figure is a slower jitter

The following figure is a slower shaking

The difference is obvious, but this difference is in the frequency domain. What if there are both high and low frequencies? As shown
in the following two pictures

This is actually two low-frequency vibrations followed by three high-frequency vibrations.

This is a high-frequency vibration three times followed by a low-frequency vibration two times

The above two actions are in opposite order, so they cannot be regarded as the same, but reflected in the frequency domain (after DFT), there seems to be no difference.
So using only FFT is not enough~~

(By the way, the vibration monitoring example in the entry-level mode is FFT, one example is training data, and the other example is comparing training data.
But it is just a threshold comparison, so it can only identify vibration and no vibration.)

Next, we need to study FSM and MLC, because they are the biggest features of LSM6DSOX, and they also have gesture recognition capabilities, so it is likely that the breakthrough point is there~

hubusbwxz69

Learned