Group Actions
Engagement
  • Author Author: dwinhold
  • Date Created: Date Created
  • Views 714 views
  • Likes 3 likes
  • Comments 5 comments
Related
Recommended

Experimenting with vibration sensors - Machinery Safety (Sort of) Blog #7

dwinhold

This is a 2 part update,

 

First, the experimenting with the sensor controlling the machine to shut off will be done in a smaller version. I will hook it up to my CNC at home. I will check the different vibrations from the big machines (but we are using 660v at work) so no electrical work for me. My CNC at home will work great for the shutdown tests.

 

Second, I had an idea this morning for another use, medical. I taped the sensor to my wrist where my pulse is, it picked up the vibrations. I attached a video of this experiment. This is another very important field the sensor can be used in. The KEMET vibration sensor has been impressing me the more I use it.

 

Dale Winhold

Attachments:
  • in reply to mcfryday

    Hi Michael, thank you for the comment (and compliment). Like I mentioned in the blogs, I never have had so much fun experimenting with this sensor as with any other. It has so many possibilities!!

  • Hi Dale, I read your blogs from the beginning and I am impressed of all the work you did ! In fact this is the most sensible sensor you can get in the market I believe and as you touched the medical area I would not be surprised that in addition to measure the heart frequency you may also be able to measure blood speed with the Doppler effect, never tried this myself but the accuracy should be far enough for this. Anyway very interesting research you did !!! Cheers, Michael

  • The windowing sampling time and start, etc, will depend on what you are trying to find... All I do is related to Jaguars and their environment protection, in this particular case, it is meant to collar on cows to avoid Jaguar attacks (which produce a retaliation killing of the Jaguars). You can train models to detect from lodger on a tree to unusual machinery vibrations. edgeImpulse has a few nice other examples about this using other methods.

     

    Cheers, have a great 2021!

    This is the result after training the neural network model:

  • Hi There, I wrote a post on that using tinyML (Machine Learning for Micro-controllers); it is meant for other uses (biodiversity care), however, as I say on the posts, this kind of things could be quite useful for preventive maintenance as well (there is plotted data as well in an unusual way which is quite effective to me), just data from an accelerometer. It is written in two parts, I hope you find this useful for what you are doing (The titles are misleading, any data that can be represented graphically can fall under the domain of machine vision).

     

    https://www.wildedge.info/post/deep-learning-machine-vision-on-mcus-introduction  and

     

    https://www.wildedge.info/post/implementing-motion-pattern-recognition-using-tinyml-on-the-sony-spresense

  • It would be interesting to plot the data.  From watching the values change in the video you can't tell if it is periodic like a pulse measurement.  Good experiment.

     

    Ralph