Hello everybody,

I'm a PhD student, meddling with model-based fault diagnosis.

Some background:
I use over-specified systems (systems with more equations than unknown variables) upon which I methodically select one equation to act as a residual generator function. If the real system is in accordance to the nominal, simulated system, then no fault has occurred.

An example:
der(x1) = -5*x1 + u;
y = x1;
where u and y are considered known and x1 is internal and hence unknown.
r = der(x1) + 5*x1 + u
  = der(y) + 5*y + u
can be selected as the residual generator function.

Notice that y acts as an input for the purposes of this use-case.

Question 1:
I know that Modelica needs the model under simulation to be just-specified (equal number of equations and unknown variables). Is this a hard constraint and a primary assumption of Modelica, or is Modelica able to propose to you the set of equations which are over-specifying the model and should be removed?

Question 2:
If the above functionality is not supported, is it something that is on some roadmap? Is it something that anyone has thought about implementing?