step_info overshoot error #554

Activity

1. bnavigator commented on Mar 2, 2021

   

   bnavigator

   on Mar 2, 2021

   Contributor

   More actions

   Thanks for bringing up the issue @juanodecc. I can confirm, although the plot is smoother with the new timevector calculation in 0.8.4 (#420), that the Overshoot value is still not what you would expect.

   

   python-control/control/timeresp.py

   Line 817 in 8cba949

   'Overshoot': 100. * (yout.max() - InfValue) / (InfValue - yout[0]),

   yout[0] here is 0.386 instead of 0. Not sure, why this is referenced here. The step_response() of a system without any further specification of initial values should start from default initial value of 0.

   @sawyerbfuller, do you think the response vector calculation should be adjusted or just the calculation of Overshoot?
2. sawyerbfuller commented on Mar 4, 2021

   

   sawyerbfuller

   on Mar 4, 2021

   Contributor

   More actions

   This is a good catch. out[0] is not zero because this is not a strictly proper transfer function: the order of the numerator and denominator are the same.

   My take on it is that the overshoot calculation has a bug that neglects this case. It should not depend on the value of yout[0].
3. sawyerbfuller

   linked a pull request that will close this issuefixed bug in calculation of percent overshoot #555on Mar 4, 2021
4. bnavigator

   closed this as completedin #555on Mar 8, 2021