How to use the state port of an integrator block in Simulink?

Hi,

I want to use the state port for initial value of the integrator when it resets. However, I get an error message when I treat it like the output port by adding it with a sumblock to equal-sized state vector, by using a delay block on the signal. The following message occurs regularly:

State ports can only be used to break algebraic loops or to "hand-off" states between systems.
Use the output port rather than the state port of 'untitled/Integrator' as the source of the signal routed (either by direct or virtual connection) to 'untitled/Sample and Hold'

Can someone elaborate on what the state port exactly is? I don't really understand from what I can find about on the internet.

Answer 1

[Jonas]

I don't really understand completely how your set-up looks like (can you post a screenshot maybe?) but I feel like instead of using the state port, you can use the output port to feed it back to the initial value port. The state port should only be used if you encounter algebraic loops when feeding the output port back to the reset or initial value input port. Standard procedure is to feed back the output port. The reason why you don't have the issue of an algebraic loop - and why you can use the output port and don't need to use the state port - is because you are using a delay block. This breaks the algebraic loop already.

The reason why the state port is used instead of the output port for algebraic loops is that these two outputs are set in a different time in the control.

Documentation info

[DB]

I want to use the state port to get the x(tk-), but I seem to only be able to use it once. The black text shows the flaws, which I want to correct because I think the state port gives the x(tk-).

[Jonas]

One thing that is wrong is that d(tk) is taken from AFTER the Unit Delay block, but it should take the signal BEFORE it.

There is going to be another mistake, but what does mean?

[DB]

Thank you for your response!

tk- means the time t just before updating/resetting at tk, with tk being the time t the system resets, every 5t. Mathematical notation would be:

[Jonas]

I understand now. This is a continuous system (hence x(t) where t is continuous time) with discrete events tk. t-k is the continuous time right before the discrete reset at tk.

Using the output of the integrator block does not suffice as x(t-k)?

[DB]

I thought the output of the integrator is x(t) as it is xdot(t) integrated?

Furthermore, wouldn't using the output of integrator as input of the initial value of the integrator lead to a algebraic loop?

[DB]

Hi @Jonas, I think I know how this should be. Could you check if I'm doing this correctly?

[Jonas]

Looks good. Does it simulate without errors?

Also, why is there *u mentioned in the Gain blocks? I would say it just applies a gain of 'A' or 'B', etc.

[DB]

Yes, no errors! :)

u is mentioned because A,B,C and are matrices and d(tk) and x(t) are vectors.