Research projects
List of current research projects of the group:
Green hydrogen is widely regarded as one of the central building blocks of a sustainable energy transition. It should be considered that hydrogen is only produced sustainably or "green" if the electricity used for water electrolysis is exclusively generated from non-fossil (i.e., renewable or regenerative) energy sources such as photovoltaics, wind, or hydropower. To ensure the long-term economic viability of water electrolysis, primarily "surplus" energy should be used for hydrogen production. This means that hydrogen should be produced in particular at times when the energy demand (over a longer period of time) is lower than the available energy potential or supply.
To accurately forecast this "surplus" energy and thus use it as efficiently as possible, precise modeling of the respective energy supply and demand is required. When modeling the supply, the geographical location of the energy source and the sometimes very strong weather-related fluctuations must be considered. Conversely, energy demand is influenced, in particular, by demographic and industrial factors. Due to the high volume of data and the volatility of various time- and space-dependent influencing variables, the energy balance can so far - if at all - only be described with very specialized, highly complex models. At the same time, due to the high complexity of these models, a meaningful, i.e., comprehensible for humans, analysis of the algorithmically made decisions is almost impossible, so no statements can be made about the reliability of the models.
This is where the project proposal comes in. Through a suitable combination, adaptation, and extension of modern methods of artificial intelligence (AI), or more precisely, machine learning (ML), various algorithms and software products (with the primary goal of modeling energy supply and demand) are to be designed and implemented. First of all, an algorithm is to be developed with the help of which the enormous quantities of, in particular, time- and space-dependent data can be reduced completely automatically to the minimum possible quantity of informative influencing variables. Subsequently, this algorithm is to be integrated into a user-friendly AI system (also to be developed) called AutoGREEN, which - following the idea of so-called Automated Machine Learning - automatically selects the most suitable algorithm possible from a set of promising ML methods, with the help of which the energy supply and the associated energy demand (at a given location) are to be modeled. Due to the focus on models that were determined based on small, but as informative as possible influencing variables, this also provides the opportunity to specifically analyze the reliability of the generated models.
This project is part of Cluster E: Hydrogen Imports from the MENA Region Compared to Hydrogen Production in Germany of the 4th Boysen-TU Dresden-Research Training Group. The focus of this highly interdisciplinary RTG is on "Hydrogen Economy -- Strategic Element of a Future Greengas Deal".
Project Duration:
since 03/2022
Contact:
© Markus Leyser 04/2023
Research Associate (Boysen-TU Dresden-Research Training Group)
NameMr Markus Leyser M.Sc.
Send encrypted email via the SecureMail portal (for TUD external users only).
Chair of Big Data Analytics in Transportation
Chair of Big Data Analytics in Transportation
Visiting address:
Bürozentrum Falkenbrunnen (FAL), Room 012 (Ground Floor) Würzburger Str. 35
01187 Dresden
The rapidly increasing availability of traffic-related data enables new potentials in terms of monitoring, analyzing, optimizing, and planning transport systems. The Mobility Data Act being pursued at the federal and EU level will further accelerate this continuous data growth. Machine learning methods, such as cluster analysis, are increasingly being used to efficiently gain valuable information and insights from large volumes of data (keyword: big data). However, the current awareness and use of corresponding algorithms in transportation science (and its applications) has so far been very limited.
This FOSTER-funded project is pursuing a comprehensive analysis and systematization of cluster analysis methods and their known applications in the context of transportation science.
Project Duration:
03/2024 -- 09/2024
Contact:
Graduate Student Assistant
NameMr Jochen Weihe B.Sc.
Send encrypted email via the SecureMail portal (for TUD external users only).
Chair of Big Data Analytics in Transportation
Chair of Big Data Analytics in Transportation
Visiting address:
Bürozentrum Falkenbrunnen (FAL), Room 120 (1st Floor) Würzburger Str. 35
01187 Dresden
This project is funded by the DAAD funding programme for project-related personal exchange (PPP) in cooperation with Darrell Whitley's group at Colorado State University.
We aim to contribute to and ideally progress the state-of-the-art research for the traveling salesperson problem (TSP), one of the most fundamental and well-studied classes of NP-hard optimization problems (Applegate et al., 2011). Apart from the research challenges that the TSP still poses itself, it also serves as the basis for related routing problems such as the Vehicle Routing Problem (VRP) or the Traveling Thief Problem (TTP). Beyond that, it is also of great importance for numerous practical applications, such as manufacturing processes, logistics, or mobility.
Due to the long history of TSP research, there exist numerous algorithms for optimizing these problems. These are typically classified into groups: exact solvers and inexact heuristics. The former tend to be slower but guarantee the optimality of their found solution (if they terminate successfully). In contrast, the inexact TSP heuristics find the optimal tours on average much faster but cannot prove the optimality of their solution.
In recent years, both research groups involved in this project have contributed to different parts of the state of the art in solving the TSP using inexact heuristics. The U.S. partners have studied the convergence behavior of TSP algorithms (Varadarajan et al., 2020) and proposed a very powerful crossover operator that is able to efficiently recombine (locally optimal) tours to produce a less costly tour (Carvalho et al., 2019). Our group, in turn, has profound knowledge in characterizing the structures of TSP instances using measurable metrics, called features (Heins et al., 2022). Further, we have conducted extensive benchmark studies, in which we compared the performances of state-of-the-art TSP heuristics (EAX and LKH in particular) across different TSP instances, and used powerful machine learning algorithms to recommend the most promising heuristic for a given instance in an automated way based on its features (Kerschke et al., 2018).
This project will now consolidate the complementary expertise of the two international research groups and thereby (a) elucidate in detail the inner mechanics of the state-of-the-art TSP heuristics, including their numerous parameterizations, (b) improve the understanding of the interactions and effects of the configurable parameters, and (c) iteratively extend both state-of-the-art heuristics to ideally develop an even better performing (and thus resource-efficient) TSP heuristic.
References
Applegate, David L., Bixby, Robert E., Chvátal, Vašek, and Cook, William J. (2011). The Traveling Salesman Problem: A Computational Study. Princeton University Press.
de Carvalho, Ozeas Quevedo, Tinós, Renato, Whitley, Darrel, and Sanches, Danilo Sipoli (2019). A New Method for Identification of Recombining Components in the Generalized Partition Crossover. 8th Brazilian Conference on Intelligent Systems (BRACIS), pp. 36-41, IEEE.
Heins, Jonathan, Bossek, Jakob, Pohl, Janina, Seiler, Moritz, Trautmann, Heike, and Kerschke, Pascal (2022). A study on the effects of normalized TSP features for automated algorithm selection. Theoretical Computer Science, pp. 123-145, Elsevier.
Kerschke, Pascal, Kotthoff, Lars, Bossek, Jakob, Hoos, Holger H., and Trautmann, Heike (2018). Leveraging TSP Solver Complementarity through Machine Learning. Evolutionary Computation, pp. 597-620, MIT Press.
Varadarajan, Swetha, Whitley, Darrell, and Ochoa, Gabriela (2020). Why Many Travelling Salesman Problem Instances Are Easier Than You Think. Proceedings of the 2020 Genetic and Evolutionary Computation Conference (GECCO), pp. 254-262, ACM.
Laufzeit des Projektes:
01/2024 – 12/2025
Ansprechpartner:
© Jonathan Heins 04/2023
Research Associate
NameMr Jonathan Heins M.Sc.
Send encrypted email via the SecureMail portal (for TUD external users only).
Chair of Big Data Analytics in Transportation
Chair of Big Data Analytics in Transportation
Visiting address:
Bürozentrum Falkenbrunnen (FAL), Room 012 (Ground Floor) Würzburger Str. 35
01187 Dresden
© Lennart Schäpermeier 04/2023
Research Associate
NameMr Lennart Schäpermeier M.Sc.
Send encrypted email via the SecureMail portal (for TUD external users only).
Chair of Big Data Analytics in Transportation
Chair of Big Data Analytics in Transportation
Visiting address:
Bürozentrum Falkenbrunnen (FAL), Room 011b (Ground Floor) Würzburger Str. 35
01187 Dresden
Office hours:
by appointment