CODEWITHSUNDEEP
labviewbest-fit-curve
UH REU
UH REU

Reputation: 1

LabView cos fitting

I am working on a program that needs to fit numerous cosine waves in order to determine one of the parameters for the function. The equation that I am using is y = y_0 + Acos((4*pi*L)/x + pi) where L is the value that I am trying to obtain from the best fit line.

I know that it is possible to do this correctly by hand for each set of data, but what is the best way to automate this process? I am currently reading in the data from text files, and running a loop with the initial paramiters changing until I have an array of paramater values that have an amplitude similar to the data, then I check the percent difference between points on the center peak and two end peaks to try to pick the best one. It in consistently picking lower values than what I get when fitting by hand (almost exactly one phase off). So is there a way to improve this method, or another method that works better?

Edit: My LabVIEW version has a cos fitting VI which is what I am using, the problem is when I try to automate the fitting by changing the initial parameters using a loop, I cant figure out how to get the program to pick the same best fit line as a human would pick.

Upvotes: 0

Views: 907

Answers (2)

CharlesB
CharlesB

Reputation: 90416

You can evaluate the goodness of the fit by computing the difference between fitting curve and your data. A VI does this in the "Advanced curve fitting" palette. Then all you have to do is pick up the best fit.

Upvotes: 0

datenwolf
datenwolf

Reputation: 162317

Why not just use a Fast Fourier Transform? This should be way faster than fitting a cosine. In the result vector of complex numbers look for the largest peak of in the totals. You're given frequency (position in the FFT result vector), amplitude and phase.

Upvotes: 2

Related Questions