Bridging the Gap Between Industrial Use Cases, Methods and Algorithms

The simulation ecosystem of the Julia programming language is rapidly growing. In order to bridge the gap between industrial use cases and methods and algorithms available in Julia, FMI.jl is released. FMI.jl aims at the integration of FMI into the Julia ecosystem by providing the possibility to load, parameterize and simulate FMUs.

Combining physical and state-of-the-art data-driven modeling inside a single simulation tool used in industry is currently not possible in a convenient fashion. An extension to FMI.jl called FMIFlux.jl allows the integration of FMUs into neural network topologies to obtain a NeuralFMU. This structural combination of a FMU (commonly used in industry) and a data-driven machine learning model combines the different advantages of both modeling approaches in a single Julia-based development environment. This allows for the usage of advanced data driven modeling techniques for physical effects that are difficult to model based on only first principles.