Tobias Heinzle

My project aims to implement a data-driven framework for optimizing maintenance and inspection actions based on records of monitoring data and past inspections by leveraging recent developments in reinforcement learning for partially observable environments.

Goals include the development of improved decision making strategies and transfer learning approaches using deep RL, taking advantage of the rich operational datasets and maintenance records provided by the Fraunhofer Institute for Wind Energy and Energy System Technology.

The resulting strategies and algorithms are to be validated at the windfarm level by means of real data over long periods of time from a considerable number of windfarms.

​

Supervisors

Daniel Straub

Eleni Chatzi

Born and raised in the Austrian state of Vorarlberg, I moved to Vienna in 2018. There I pursued a Bachelor and subsequent Master in Mathematics at the University of Vienna, which I finished in March 2025.

During my studies I taught programming and tutorial classes, and in 2022 got hired as a software developer at IBM Austria. There I stayed until March 2025.

Additionally, I worked as a research assistant at the Institute of Astrophysics in Vienna throughout the summer of 2024, which gave me a first glimpse into the world of research.

These experiences, as well as the process of writing a master thesis, led me to pursue a PhD, particularly in the area of data-driven optimal planning and decision strategies.

Motivation

The research proposal of IntelliWind intrigued me because of its extensive real world impact and the interesting computational challenges involved.

My background in applied mathematics and software engineering, paired with my strong interest in decision 

making algorithms and reinforcement learning, very closely aligns with the goals and topics of the project.

The application in the domain of sustainable energy also resonates with my personal values, and I feel very privileged to contribute in this area.