Question: Setting FMU inputs in fmi2EvaluateME()

[hejersbo]

Hi,

This is not a bug report. Merely a question (and perhaps clarification of a my misunderstanding :-))

I would like to specify the FMU inputs when calling fmiEvaluteME() and I can see that the fmiEvaluate() has arguments setValueReferences and setValues. But is it possible to set an input function (like for fmiSimulateME) that is evaluated with the input time of the ODE solver. So e.g. one could have something like sin(t) being the input function for the FMU.

[hejersbo]

The reason for this is that I try to make a simple example - more or less replicating the simple_hybrid_ME example of the repo but with a different ME FMU. But I setup the neural FMU just like the example (using fmiEvaluateME()).

# NeuralFMU setup
numStates = fmiGetNumberOfStates(referenceFMU)

net = Chain(inputs -> fmiEvaluateME(referenceFMU, inputs),
            Dense(numStates, 16, tanh),
            Dense(16, 16, tanh),
            Dense(16, numStates))

neuralFMU = ME_NeuralFMU(referenceFMU, net, (tStart, tStop), Tsit5(); saveat=tSave);

[ThummeTo]

Sorry for the late reply, somehow I skipped that notification...

Yes, you can totally do that, see this test here:
https://github.com/ThummeTo/FMIFlux.jl/blob/568e585d607abe25e51b20a5c7fadb7be38807fa/test/hybrid_ME.jl#LL134C6-L134C6

Basically your chain would look like this:

function myInputFunction(t)
   return [sin(t)]
end

net = Chain(x-> referenceFMU(;x=x, u_refs=setVRs, u=myInputFunction(  getCurrentComponent(referenceFMU).t  )),
            Dense(numStates, 16, tanh),
            Dense(16, 16, tanh),
            Dense(16, numStates))

getCurrentComponent(referenceFMU).t equals the time instance of the ODE solver.