1. Home
  2. ⚙ Mechanical Engineering
  3. Double Integration. Acceleration to Position

Double Integration. Acceleration to Position

Dear All, I have some data from a force plate (a person jumping up and down

10times...so the data has peaks when the person is pushing off of the force plate and the forceplate reads 0 when the person is in the air) and would like to go from this data to position of the body jumping on the force plate. I am using the following code in Matlab, but having what appears to be strange results:

%GRF is in Newtons - column vector %Mass - kg - single value %Sampling Rate of force plate = 1500Hz - single value

Acceleration = (GRF/Mass) - 9.81; Velocity = (1/1500) * cumtrapz(Acceleration); Position = (1/1500) * cumtrapz(Velocity);

I think using the cumtrapz function is my problem??? That is, when I plot velocity versus time, it looks like a sine wave but the sine wave is always increasing (it seems to increase without returning to zero...looks like multiple sine waves on a slant). I think if I plot this for one jump, the data looks correct, but for the entire trial it does not seem to be working. Is this correct? Any help would be greatly appreciated.

Reply to
jmp
Loading thread data ...

I don't think you can get there from here........you have F(t) force as a function of time

the person jumping is an elastic system as is your structure on which the force plate (load cell) sits.

Knowing F(t) is a start but you need a dynamic system model because F=ma works for rigid body systems

you have F(t)= mX'' + cX' + kX

cheers Bob

Reply to
BobK207

"jmp" wrote in news:1163127282.972971.264790 @e3g2000cwe.googlegroups.com:

You need to use gravity as a constant resisted by the force on the plate.

That is m*a=m*g-Fplate

Reply to
Greg Locock

I'm not sure, but I'm betting the problem is a small offset, or bias, in the force plate data. I've put an accelerometer on a piston in an air compressor (head off) and run it at several speeds. When you double-integrate it, you *never* get a periodic position function. Small offsets in the original data translate to ramps in velocity and parabolae in position. I'm guessing your results in velocity look like a periodic function added to a ramp. In position, the parabola overwhelms the data.

You might be able to salvage something by tweaking the grav constant (since that looks like a constant offset to the force plate.

On second thought,the best approach would be to add a step at the beginning where you add/subtract a small constant to the FP data, but you'll need an iterative approach to find out what it is. You'll know you get there when the position function is bounded.

Hope this helps. JFG

Reply to
jfg4

PolyTech Forum website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.