Event Isolation – Slope not picking up all events for me

[norbert.bakkers@lr.org]

I’m trying to remove steps in a signal using the event isolation option.
By setting a slope value (positive and negative)higher than a naturally expected value I can assume that all selected cases are steps in the signal. However, many of the steps higher than the slope value are not picked up by the event isolation routine.
Am I doing something wrong here?
I’ve attached a sample file indicating my problem.

[norbert.bakkers@lr.org]

I’m trying to remove steps in a signal using the event isolation option.
By setting a slope value (positive and negative)higher than a naturally expected value I can assume that all selected cases are steps in the signal. However, many of the steps higher than the slope value are not picked up by the event isolation routine.
Am I doing something wrong here?
I’ve attached a sample file indicating my problem.

[Bernhard Kantz]

The reason could be a rounding problem while calculating the delta x, means that the calculated delta x is greater than 0.04.

Therefore you should use a value greater than 0.04 for the delta-x in the event isolation object, e.g. 0.041.

support@weisang.com

[norbert.bakkers@lr.org]

Increasing the dX (delta X) to 0.41 works.
However, in my understanding, the slope should be purely defined by the given dX and dY (slope = dY/dX). This means that a rounding error in the dX should not have such a profound effect on the obtained results.
For my example the slope threshold should be 0.001/0.4=0.0025 but I have cases with a slope of 0.2655 which are not selected just because the dX was 0.400085.

[Bernhard Kantz]

You are right.
The current algorithm doesn’t find an event if there is only one value in the specified delta x range.
We will improve the algorithm for this case in a future FlexPro release.

support@weisang.com

[norbert.bakkers@lr.org]

Thank you.
I will use a dX greater than the sample time for the time being.

[Author]

Posts