Detect Slope change from noisy data

[tars364]

Hi Guys I have data, which is noisy but linear( you can find the data in the VI attached), is it possible to find the location where the slope is changed or any other ideas are welcome?

P.S My data can be negative or positive under different conditions, so I couldn't use comparison tools here.

Thanks

[GerdW · Accepted solution]

Hi tars,

see this:

You can detect that outlier quite easily I guess!?

Btw. never compare floats for (in)equality! Most often the result might be wrong…

[tars364 · Accepted solution]

Hi Gred, 

Thanks for the reply. NIce idea

Best Regards,

Tarun

[HYDAC_R_and_D_Lab]

Hello,

I am also looking to detect the slope change with data from a "noise" source, pressure sensor. The signal is not as clean as the demo signal for detecting a change in slope, so when I saw this it interested me to see if I could apply it to my applications. This seems to detect the change in slope at the time axis, correct?

I would like to detect the change in slope and have the amplitude reported (y-axis value). Any thoughts on this? 

I attached the sample vi with my sample data file. 

[GerdW]

Hi Hydac,

even with noise there is a clear peak in the slope of your signal:

Unfortunately I cannot open your VI (it would help to downconvert to LV2021 or LV2019), so I cannot comment on your code…

[HYDAC_R_and_D_Lab]

Try this version. I think I saved it correctly to a past version of LabView.

Thanks, 

[HYDAC_R_and_D_Lab]

That "knee" in the curve is what I am after. Of course I can measure this after the data is saved in Excel for example. However I am developing a process where I can detect this as we run the test, or at lease write the data, then run the analysis in LabView on the spot.

Thanks for looking!