time-based data to length-based data

[yuanhui]

For a rolling process, normally the time-based data (say, mill speed at each sampling time) is used to do some analysis. Now I wanna do some further analysis using the length-based data. Again, for mill speed, I want the data at different lengths instead of sampling time. Note that length-based data is unequally-spaced. I wonder whether in LabView there are some modules to do this kind of resampling. Thanks for any suggestion.

[Hummer1]

If you multiply the time by the velocities, you could add a length value as a column. 

For each time, take the mill speed and multiply it by the elapsed time since the last speed sample...that should  give you the distance the mill has traveled between the samples...you could do some neat stuff like take the speed at time1 and time2 and average them and use that as the speed of the mill between the samples...(I suspect you are sampeling the speed much faster than it is changing).

Does that help?

Hummer1

[yuanhui]

Above all, thanks for your idea. Actually, I thought about that before. Just there is one problem. The data is sampled at 1ms. Using this kind of integration idea, the computation will takes a while even for few-minute data.

For mill speed, it has an acceleration process at the beginning and a decelaration one at the end. I am also wondering the analysis accuracy if using the average speed.

That is why I am thinking whether LabView has such a VI. 

[Hummer1]

If the sampeling is much faster than the changes in the speed of the line, then you could use some averaging on some of the speed numbers to smooth the data.  You could "window" or use a "running average" to calculate the distance.  Why don't you build a little simulation vi using different sample rates and distances and change rates being random variables and see how different techniques handle the problem.

Each time you make an estimate of the speed average multiplied by the time, you could accumulate the distance...you could keep a running total of the distance at each average time...that shouldn't take too much time to calculate...even if you did it for say 10 or 100ms increments...It all depends on how fast the physical line is running and how much inertia it has ... (i.e. can it change speed quickly...you could even find that out by measuring the differences in velocity over time and calculating the acceleration available to the system..., but now we are into the F=ma stuff...)

Good Luck...