We would like to introduce you to the methods and tools we use for our research, coaching and teaching. At optiflow we deal with the application and development of 3D models, numerical grids and simulations for different engineering applications. In particular, we are very interested in complex physical or chemically reactive systems or the motion and flow in particle systems. We are working on our in-house developed artificial intelligence (AI) software tool for collaboration with digital twins. The optiflowNeuralNetwort (ONN) tool. Currently we work on the Evaluation and training of our ONN with data from river beds under the Framework of the ZUG - Zukunft, Umwelt Gesellschaft (KI-Leuchttürme für Umwelt, Klima, Natur und Ressourcen; https://www.z-u-g.org/foerderung/ki-leuchttuerme-fuer-umwelt-klima-natur-und-ressourcen/ ).
The proprietary AI analysis tool optiflowNeuralNetwork (ONN) was developed at htw saar over several years. It is based on Fahlman's Quickprop program translated from Common Lisp into C by Regier at the University of California, Berkeley [1]. Quickprop is an almost forgotten neural training algorithm also mentioned by Bonaccorso [2]. The base version is Quickprop was from 1988. We developed ist further and work now with an version in Qt/C++ and use it for the training of our river bed data. An example of network setup data and raw measurement data of the river bed of the Saar river is provided here. is included at the end of this file. The algorithm and some test results are described in Fahlman [1].
You can download the latest release of our artificial intelligence (AI) software for Microsoft Windows. The optiflowNeuralNetwort (ONN) tool is written in Qt6. The examples "frog_example.net" or "river_example.net" can be found inside the installation folder of the ONN. You can also find the raw data from the boat measurements there.
Windows: Please, click here to download the latest version of optiflowNeuralNetwort (ONN) v1.12 and install it ...
Here you can download an Blender file (*.blend) for the postprocessing and visualisation of the river bed data.
Example_RiverBed.blend
For advanced simulations like, e.g. computational fluid dynamics (CFD) we use the software tools ANSYS CFX, ANSYS Chemkin, ANSYS Fluent, openFOAM or Simcenter Amesim. We also work on geometry creation and discretization with Blender, Unreal Engine 5 and Unity 6. We can generate hexahedral lattices with ICEM or blockmesh. For blockmesh we created an automatic tool for faster meshing. Our partners benefit from our many years of experience in the industrial field of modeling and simulation as well as from our network in the scientific field and the latest results from our research projects.
We work on software solutions for Qt and Android, e.g. Grill Simulator
If you have any question, please send us an E-mail: frank.rueckert@htwsaar.de
- [1] Scott E. Fahlman: Faster-Learning Variations on Back-Propagation: An Empirical Study; in Proceedings of 1988 Connectionist Models Summer School, published Morgan Kaufmann
- [2] Matteo Bonotto, A.C.: Krylov subspace methods for model order reduction in computational electromagnetics; IFAC-PapersOnLine (pp. 6355-6360). Elsevier, 2017.
