COMET-Project „TransMet1” started

HEAL is part of the COMET IC-MPPE project “TransMet1”, which is dedicated to the transformation in metallurgy to recycled steel. The objective is to provide fundamentals and tools for the production of high-quality recycled and CO2-reduced strip steels.

Based on two use cases, a demonstrator of a material and process design and optimization software based on hybrid modeling techniques for offline use in steel sheet production will be established. Hybrid modelling combines physics-based models with machine learning models. The main strategy is to use a combination of physical and data-driven modeling approaches for the development of process planning tools for steel strip production.

Use case 1 focuses on hybrid model-based prediction of the phase transformation of well-recrystallized austenite into its child phases. 

Use case 2 is devoted to hybrid model-based prediction of the deformation behavior. 

The task of HEAL in the project is to develop symbolic regression algorithms that can be used for hybrid modeling in use case 1. 

Important impacts of the project are the reduction of CO2 emissions due to a switch from the blast furnace route to an electric arc furnace route, an economical use of raw materials such as iron ore and coal due to a radical switch to circular economy and a more efficient steel strip production with lower rejection rates.

New Project „In Process Optimisation (IPO)”

HEAL supports GAMED ( in Graz, Austria in the development of self-learning quality control software for the food industry. The new collaborative project supported by the Austrian Research Promotion Agency (FFG) aims to use KI-driven in process optimisation (IPO) to improve product safety, product quality and cost-effectiveness.

The software solution will allow centralized access to information from different departments, and process flows and provide the basis to reduce throughput times, manufacturing costs and rejection rates. In the event of deviations, it will become possible to immediately adapt test plans, which leads to a reduction in customer reclamation rates and recalls and thus strengthens international competitiveness.

HEAL supports the implementation and integration of machine learning components and advises on the selection and application of suitable algorithms learning tasks.

Mitarbeiter*innen gesucht!

Im Rahmen eines FWF Projektes oder eines Josef Ressel Zentrums können Sie in der Forschungsgruppe HEAL mitarbeiten (2 Vollzeitstellen): 


  • Mitarbeit in F&E-Projekten im Bereich KI mit Schwerpunkt maschinelles Lernen und Optimierung
  • Anwendung, Erweiterung und Entwicklung von Methoden der dynamischen Optimierung
  • Umsetzung wissenschaftlicher Konzepte für praxisrelevante Aufgabenstellungen
  • Publikationstätigkeit und Erstellung neuer Forschungsanträge


  • Abgeschlossenes technisches Studium auf Masterniveau (z.B. Informatik, Mathematik, …)
  • Interesse an heuristischer Optimierung, evolutionären Algorithmen und maschinellem Lernen
  • Erfahrung in der Softwareentwicklung (bevorzugt .NET und C#)
  • Möglichkeit zur Dissertation an einer unserer Partner-Universitäten
  • Intrinsische Motivation zur wissenschaftlichen Arbeit
  • Multidisziplinäre Herangehensweise und offene Umgangsformen
  • Verlässlichkeit, Flexibilität und hohe Einsatzbereitschaft
  • Sehr gute Deutsch- und Englischkenntnisse

Wir freuen uns auf Ihre Bewerbung! / We are looking forward to your application!

FH-Prof. PD DI Dr. Michael Affenzeller

Farewell to Eva Maria Holzleitner and Viktoria Hauder

It is with a heavy heart that we said goodbye to Eva-Maria Holzleitner and Viktoria Hauder from the HEAL research group today.  Thank you very much for your energy, friendship and support.

Your next stations are enormously exciting, responsible and challenging. Therefore, the smiling eye prevails, because we will observe, support and sometimes hopefully accompany your ways.

All the best

The HEALers

1st HeuristicLab Community Meeting

Our open source software HeuristicLab - used for optimization and machine learning - has been in development since 2002 and had mainly international users in the past. In recent years we have observed that a growing group (especially at the JKU, FH and non-university research institutions) is forming here as well, which is modeling and optimizing with HeuristicLab or sees potential in doing this more intensively in the future.

From this background we started a first event, to bring together the groups of HeuristicLab users, developers and those yet to become one for a constructive exchange and friendly get-together. The first event of this kind took place on June 21st, 2021.

Head of the research group HEAL, Michael Affenzeller, gave a first input on research group activities, followed by Stefan Wagner (HeuristicLab), Andreas Beham (optimization of production and logistics) and Gabriel Kronberger (data analytics and machine learning).

After that, Wolfgang Roland from the Institute for Polymer Extrusion and Compounding (IPEC) and Siegfried Silber from the Linz Center of Mechatronics (LCM), talked about their experience using HeuristicLab.

Finally, Stephan Winkler lead through a panel discussion with Stefan Wagner, Wolfgang Roland, Robert Wille, Siegfried Silber and Michael Affenzeller to talk about the future development of the software.


Subscribe to News