Skip to content

named transfer functions #885

Closed
Closed

Description

@mzaffalon

Would you consider having named transfer functions and perhaps even moving named_ss from RobustAndOptimalControl.jl to ControlSystems.jl?

I am asking because the bodeplot gives generic names to the outputs and because for LTISystems, the default name is empty

Activity

  1. baggepinnen commented on Oct 12, 2023

    @baggepinnen
    Member

    I have considered implementing a new StateSpace type that includes a lot of the functionality that is currently scattered across multiple different statespace types, such as

    • Named signals
    • Delays
    • Uncertainties
    • Static nonlinearities in feedback path

    Unfortunately, moving the implementation of a function from ROC to CS cannot be done in a non-breaking way if we want to export the function from CS. I'm also hesitant bringing all that code into ControlSystems, the NamedStateSpace type is still quite immature with lots of problems in how it interacts with other system types and functions not handling it properly etc. I'm thus not yet ready to bring it here and signal that it's ready and stabe. RobustAndOptimalControl is still at version 0.x, which accurately signals that it's not quite as mature as ContorlSystems is.

    What exactly is the problem with bodeplot, you should be able to set the names yourself using the regular Plots keywords?

    As for named transfer functions, implementing such a type would be a ton of effort, the incomplete implementation of NamedStateSpace is already very long, and it still has a lot of gaps. Do you have a use case for which converting your transfer function to a statespace system wouldn't work?

  2. mzaffalon commented on Oct 12, 2023

    @mzaffalon
    ContributorAuthor

    I understand.

    No particular application in mind except for having the plots automatically add the label.

    I also liked the ease with which one can build complicated connections used named_ss and wondered why one has to import an extra package for something that would ideally belong to ControlSystems.jl or even ControlSystemsBase.jl.

  3. baggepinnen commented on Oct 12, 2023

    @baggepinnen
    Member

    Yeah, the reason there is simply that I don't want to put experimental and immature features in a package that aims at being stable and mature :)

    Control engineers are often risk averse (for good reasons, we don't want planes falling out of the sky), so trying hard to signal stability and maturity is especially important for CS.jl.

  4. baggepinnen commented on Oct 12, 2023

    @baggepinnen
    Member

    I should mention that there is nothing preventing you from implementing a package NamedTransferFuncitons.jl, as long as you extend the appropriate functions

    • ControlSystemsBase.state_names
    • ControlSystemsBase.input_names
    • ControlSystemsBase.output_names

    The bodeplot will work as you want

  5. mzaffalon commented on Oct 12, 2023

    @mzaffalon
    ContributorAuthor

    Ah, and the other reason was that RobustAndOptimalControl requires anyway ControlSystems. That said, you are the best person with an overview of the packages development.

    As for you other comment: one should also extend TransferFunction to have entries for system, inputs and outputs names.

  6. locked and limited conversation to collaborators on Oct 12, 2023
  7. converted this issue into a discussion #886 on Oct 12, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions