Data Filtering

I finally have obtained force plate data with the use of the MATLAB data aquisition toolbox. I think the data contain a lot of noise. With that, I want to know how to filter the data.

Thank you

Answer 1

[Walter Roberson]

You need a theoretical model of the noise in order to filter it properly. If you don't have a theoretical model, then I could say, "Just zero all of the data" and that would be a perfectly smooth solution that you wouldn't be able to discount.

[Sangwoo]

Thank you for quick response.

What does " theoretical model" mean?? As a matter of fact, I am planning to use the 4th order butterworth lowpass filter.

[Walter Roberson]

For example, how much of the noise is in the item being measured, and how much of the noise is in the technology used for doing the measurements?

To toss some things out randomly: are the force measurements affected by frictive heating and if so how does that heat dissipate as a function of time and distance? Is the force being measured indirectly by optical fringing, and if so then does that imply you have quantum noise problems because of the use of CMOS or CCD detectors?

[Walter Roberson]

Is whatever sensing being done linear response or non-linear?

Is the noise additive noise or multiplicative noise? Is some of the noise due to quantum uncertainty "smearing" results?