PyFMI | Functional Mock-Up Units both for Model Exchange | Mock library

I had some success in deploying my machine learning model (already trained) in a simulation environment (OpenModelica, in this context) via an external C-function.

However, to standardise the process, I am aiming to use the FMI standard instead of the external C-function. Therefore I need to wrap my ML model as an FMU.

Is there any way that I can do that? I read about PyFMI, however, it seems that it only controls the Co-simulation in a Python environment, instead of wrapping your ML model as an FMU.

The goal is to produce FMU from a trained ML model, and then deploy this FMU in a simulation environment (OpenModelica, for example). Any help will be very much appreciated.

Thanks

With Python libraries like FMPy I am able to simulate fmus (using fmpy.simulate_fmu) for given start_time and stop_time. In such case, the function simulate_fmu completes the simulation and return the time-series results.

However, I want to create a closed loop between fmu and a Python function (i.e. in a Python script initialize the fmu, get the results from fmu after every 0.1s and based on that update the input value to the fmu for the next timestep). Is there a way to achieve this using existing libraries like fmpy or pyfmi?

I have started to in PyFMI use parameter estimation with the procedure model.estimate() and works well.

From the documentation (Andersson et al 2016) as well as practical use I understand that model parameters are taken from the compiled FMU-model if not estimated. It would have been very practical to have an option to provide a dictionary with a set of the fixed parameter values different from the default of the model. Is there any way to provide that?

The current workflow is that for a larger model built up of parts from libraries, then you need to make a copy of these models and set parameters to the proper value in the code, and then compile it. It is a somewhat tedious procedure. Perhaps I have misunderstood something?

Andersson et al (2016): "PyFMI: A Python package for…” https://portal.research.lu.se/portal/files/7201641/pyfmi_tech.pdf

I have a FMU of a model and the use case is to change parameter values of the FMU to see the impact on the results. Is there a way to list top level parameters of the FMU using either FMPy or pyFMI if I dont' have access to the Modelica model?

One of the process I have been following is to open the FMU using FMPy.gui and go through the list of parameters and then use them in the script but I would like to know if there an easier way of doing it so that I can list then in the Jupyter notebook and change the parameters as needed?

I would like to install Python 3 with PyFMI at my Windows computer where I already have JModelica 2.10 with Python 2. My idea is thus to compile Modelica models to FMUs in Python 2 as necessary but then develop scripts in Python 3 to run the FMUs and visualise results. How do I best do this to avoid any conflict between the two Python environments?

My goal is to be able to run FMUs produced by OpenModelica in Ubuntu 18.04 and then run these with PyFMI in Python 3 environment.

I follow the outline for PyFMI installation here https://jmodelica.org/pyfmi/installation.html.

So far I have using Conda managed to install Python3, Numpy, Scipy, lxml and some other packages and made it work with some of my Python examples. But I would appreciate some detailed advice how to

1. Install FMI Library - and I do not get how to setup the flag fmil-home
2. Install Assimulo

After that I guess we are ready to do from the installation outline “python setup.py install —fmil-home=/path/to/fmil"

Appreciate some basic advice!

When simulating the cartpole problem as an FMU in PyFMI I get different results giving the same input depending on if I use "model exchange" or "co-simulation". The ME results are correct, the CS results seem completely off.

Model exchange results