Use of Matlab Function in State Space Modeling gives Dimension Errors - S-Function needed?

Dear Community,

I am modeling a heat transfer problem using a state space model. For that I used the Matlab Function block, which looks approximately like this:

function [x1p, x2p, x3p] = fcn(x1, x2, x3, u1, u2, u3, u4, u5, u6, u7, u8, u11, u12)
parameter1 = ...;
parameter2 = ...;
parameter3 = ...;
etc.
[...]

x1p = ...; %ode1
x2p = ...; %ode2
x3p = ...; %ode3

The list of parameters is quite long. What I want to do is exclude them from the matlab function and initialize them once via the accompanying script. Apparently because of the loop that is necessary (see the attached picture), I get dimension errors, when I do delete the constant parameters in the matlab function and initialize them from my script.

What do I need? Is it an S-function, or a Level-2 Matlab S function? I have zero experience with those, but I found via the search that the Matlab function treats its parameters as "persistent variables", in other words local to the function, if I understand correctly.

Can you give me any advice on how to treat this problem?

Thanks in advance

Felix

Edit: Error message looks like this:

Simulation

1

Clear

05:16 PM Elapsed: 1 sec

Compilation of model 'test' failed while trying to resolve underspecified signal dimensions.

Suggested Actions

Enable 'warning' or 'error' diagnostics for the list of underspecified signal dimensions.Open

Caused by:

• Error in default port dimensions function of S-function 'test_sfunc/Subsystem/Matlab Function'. This function does not fully set the dimensions of output port 2

[Felix Schönig]

*bump*

[darova]

I have no knowledge about Simulink. This might be usefull

function xp = fcn(x,u) %x1, x2, x3, u1, u2, u3, u4, u5, u6, u7, u8, u11, u12)
% a = u(1);
% b = u(2);
% ...
parameter1 = ...;
parameter2 = ...;
parameter3 = ...;
etc.
[...]

xp = zeros(3,1);
xp(1) = ...; %ode1
xp(2) = ...; %ode2
xp(3) = ...; %ode3

Usually if you have something to do with ode solvers. The function header looks like this

function xp = fcn(t,x,parameters)    % time, x variable, parameters

• What I want to do is exclude them from the matlab function and initialize them once via the accompanying script

If your parameters don't change during simulation you don't need persistent

[Felix Schönig]

Hello darova, thanks for replying. Indeed the parameters dont change during the simulation. However as there are many identical (function) blocks chained to another I want to keep the functions arguments at an absolute minimum, thus excluding the parameters. Like this:

function xp = fcn(x,u)

*NOTHING*

xp = parameter1 * x + parameter2 * u^2; %for example

But apparently the parameters can not be initialized separately in a script and then used from the workspace but have to be initialized inside the function body. Why is that? I have also tried loading them inside the function from a .mat-file and declaring them as global. It didn´t help.

[darova]

If you have .mat file with parameters. Try this

function xp = fcn(x,u)
S = load('parameters.mat');     % load parameters into structure
par1 = S.par1;
par2 = S.par2;
% ...

Wher you are using dynamic command

load parameters.mat

It tries to load it to yourmain workspace. But doesn't do this into function workspace

[Felix Schönig]

Thank you! Even though that is not the solution I had wished for (without having to initialize another list of parameters with par1 = ... par2 = ...), it sounds like a good workaround to centralize my parameter list via the .mat file. I will try it tomorrow. :) Thanks so much and have a good evening!

[darova]

There is one more way (but is not recommended): using global variables. See of one of topics: LINK

somewhere in main script

global a b c
a = 2;
b = 3;
c = 4;

then to use them in a function

function dx = f(t,x)
global a                % don't forget declaring
dx(1) = a*x(1);
end