Android Question Accelerometer Useage

[Harris]

Does anyone have experience in using the accelerometer for detecting and determining vibration?

Here is my use case.
Motor Grader.
This machine grades the road surface to remove ice, snow or to level a rutted road.
In order to determine if the machine is actually doing useful work, I would like to sense the vibration of when the blade is in contact with the road surface and producing significantly more vibs than when the blade is up and travelling.

This app would be a test to determine the difference between the two modes but what would I expect to see from the sensor (values) and what would the x, y , z represent?

Is there a way to control the rate at which data is published, or would using a timer to sample in-coming data be best to control the amount of "noise"?

Thanks

 

[JTmartins]

Hi Harris,

I've played with accelerometer a little for a similar purpose..And I can tell you, that according to my tests, unless you are traveling in a very smooth road, it is almost impossible to determne anything significant. Even a smal stone on the road or a micro bump or hole, will make the accelerometer values skyroof as if you where falling from a plane.

I think it will be extremly difficult to achive what you want with the device acelerometer. Ive tryed all kind of smoothing filters on data to eliminate bumps, or small road depressions, and although a little improvement could be seen, I've never achieved any results that I could consider really usefull.

In other words, the accelerometer is extremly sensible, probably to sensible for your purpose...But...

 

[Erel]

Check this tutorial: https://www.b4x.com/android/forum/threads/android-shake-event-with-phonesensors.9857/#content

 

[Harris]

Hi Harris,

I've played with accelerometer a little for a similar purpose..And I can tell you, that according to my tests, unless you are traveling in a very smooth road, it is almost impossible to determne anything significant. Even a smal stone on the road or a micro bump or hole, will make the accelerometer values skyroof as if you where falling from a plane.

I think it will be extremly difficult to achive what you want with the device acelerometer. Ive tryed all kind of smoothing filters on data to eliminate bumps, or small road depressions, and although a little improvement could be seen, I've never achieved any results that I could consider really usefull.

In other words, the accelerometer is extremley sensible, probably to sensible for your purpose...But...

Thanks JT,

I think I see what you mean.

The MaxValue of the acc. sensor is 19.6...
With a rough road, or engine running, or blade scraping - it seems (from my desk anyways) that it wouldn't be difficult to pin these values (x & y) just sitting and idling - let alone scraping hard pack. The Android device accellerometer sensor it just too sensitive!

So, with that, how does one determine the difference in modes when x and y are constantly at Max?

What would be nice is the ability to adjust the max values (or sensitivity) so one could "tune" the upper and lower frequency that one needs for a specific task. Let's say 0 - 100 for example. 0 - no vib, 100 being max vib (8.0 on the ricter scale). The lower and upper limits would have to be set (hardware level) to one's environment.

To accomplish this properly (I suppose - chime in here anyone), one needs an external g-force sensor (probably a legacy serial RS-232) that is adjustable for the mix / max vibration found thru testing on a machine - hook it to a serial BlueTooth and get the data in this fashion.
Better yet, mount the g-force sensor on the blade and run a wire back to the controller. It IS this attachment we really want to know about...

There is definitely a market for this device and data...

Hello Google, what did you find in this regard?

Thanks

 

[JTmartins]

You can try to use the device acceleromenter and making a log application, as in the post Erel pointed out.

Do some tests...like 2 minutes with blade on, 2 minutes with blade off in a multitude of different road environments.

Make a note of the changes, so you can better interpret data.

Read the data in an Excell spreadsheet...Make some graphs, and see if you can extract some meaningfull information.

You can try some filtering on data to smooth it using things like mathlab. Do a search for data smoothing algorithms.

Maybe your're on to something. We never know, until we try.

In my case, it didn't work out as even a car passing over a metal santation cover could be noticed, and when going trough bad roads (very common in my country) the data was pretty useless. However It did work if we where travelling in a motor way.

I have some doubts, but as I said...try out, and see how it goes.

That's science