extrapolation of a measured curve

[Kejoglo]

Hello all,

I'll try and be specific with this extrapolation problem.

We have a curve which is calculated from a series of measurements. We typically end up with a curve of about 1000 points and perhaps the last 60 or 70 or so are incorrect (think of them as being obscured). As we always know where this happens we can approximate by taking point on the curve where we know we have a good reading and then fit a curve in order to extrapolate to the end of the actual measurement (basically extrapolate over the obscuation).

In the past older (non-labVIEW) software has used a quadratic fit to extrapolate. Should I be using the Curve fitting vi and if so are there any examples of this kind of problem? Has anyone any ideas.

Any shared experience or know-how would be appreciated.

Regards

Kevin

[DFGray]

In general, Express VIs are useful to try out new algorithms and quickly get you started.  If you have an established, working algorithm, there is no real need to change it.  Express VIs tend to have more overhead than hand coded routines, so if you are size or speed limited, that is something to take into account (it doesn't sound like you would have a problem).  However, as I said in the first sentence, the curve fitting Express VI is a great way to try out some algorithms you haven't tried before.  For example, if you know your data model is actually more complex than a quadratic, the curve fit Express VI will help you find a more suitable fit and can show you how to code it in LabVIEW, if you want to go back to a custom analysis routine.

Take home message - if you are satisfied with your current analysis, there is no reason to change.  If you want to try something new, give it a try.