Systems and Control Seminar
2024/05/30: Prof. Dr. Tobias Breiten, Technische Universität Berlin
- 2024/05/30: Prof. Dr. Tobias Breiten (TU Berlin) will give a talk on An approach to nonlinear observer design via optimal control theory – the Mortensen observer at 4:00 pm in 2/W014 (in conjunction with Chemnitzer Mathematisches Colloquium).
- Abstract: State observers for dynamical systems have a rich history in research and application. For linear systems the Kalman filter offers a practically feasible and theoretically well understood solution. For non linear systems, however, research is still ongoing. In this talk we present the minimum energy estimator, proposed among others by R. E. Mortensen in 1968. In particular we discuss theoretical results on well-posedness and propose a scheme for its numerical realization.
- Curriculum Vitae: Tobias Breiten studied in Kaiserslautern where he received his Diploma degree in Technomathematik in 2009. He started his PhD in the group of Peter Benner in Chemnitz which 2010 jointly moved to the Max Planck Institute in Magdeburg. As a member of the International Max Planck Research School, Tobias received his PhD from the Otto-von-Guericke Universität in 2013. In his dissertation, he discussed model reduction methods for nonlinear control systems for which he was awarded the Otto Hahn Medal from the Max Planck Society in 2014. He then moved to the University of Graz where he obtained his habilitation in 2018 (supervised by Karl Kunisch). In 2020 he took up a professor position at Technische Universit ät Berlin. His current research interests include model reduction, (infinite-dimensional) control theory, PDE constrained optimization and nonlinear observer design.
2024/04/24: Prof. Dr.-Ing. Alexander Schaum, University of Hohenheim
- 2024/04/24: Prof. Dr.-Ing. Alexander Schaum (University of Hohenheim) will give a talk on Monitoring and control of balance models with focus on bio- and food process engineering at 1:15 pm in 2/W065.
- Abstract: Balance equations are a fundamental tool for modeling of process engineering systems. They provide efficient means for addressing different areas of application on a common structural basis provided by fundamental physical laws like mass and energy conservation. In the talk, basic concepts about balance equations will be reviewed in the context of bio- and food engineering applications and beyond. Consequences from their structural properties on their dynamic behavior will be used for the design of monitoring and control strategies. In this regard particular focus will be put on the design of nonlinear observers for such systems. Observer design is of particular importance in these application areas as bio- and food engineering systems always are determined by a large number of internal states that are not directly measurable but which are essential for the system behavior in regard of product quality and safety, and thus need to be monitored during the running process. Furthermore, as biological system are always subject to considerable variabilities, robustness properties need to be analyzed carefully. These aspects will be discussed based on some illustrative examples, involving cell populations and thermal processing.
- Curriculum Vitae: Alexander Schaum holds the chair on Process Analytics since September 2023 at the University of Hohenheim. He received his Diploma degree in Technical Cybernetics in 2006 from the University of Stuttgart, his PhD in Control Engineering in 2009 from Universidad Nacional Autónoma de México (UNAM) in Mexico City and his habilitation degree in Control Theory and Control Engineering in 2020 from Kiel University (CAU). From 2011 to 2014 he was a guest professor with the Department for Applied Mathematics and Systems at the Universidad Autonoma Metropolitana (UAM) in Cuajmalpa, Mexico City. During winter term 2018 he had an interim professorship for Networked Electronic Systems and in summer term 2023 for Automation and Control, both at CAU. He is IEEE Senior Member, IFAC Affiliate, member of the IEEE Control System Society (CSS), DECHEMA, GDCH, and ARGESIM. His research interests focus on automation, process monitoring and control with applications in Bio- and Food Technology, thermal processing, as well as development of new methods for observer and control design for nonlinear, distributed parameter and stochastic systems.
2024/04/11: Prof. Dr. Vedran S. Perić, Aarhus University
- 2024/04/11: Prof. Dr. Vedran S. Perić (Aarhus University, Denmark) will give a talk on Revealing Data-driven Model Uncertainty: Leveraging the Prediction Error System Identification Theory at 09:30 pm in 2/W040.
- Abstract: Data-driven modeling methods have garnered significant attention from researchers across various fields. Among these approaches, the Prediction Error System Identification Method stands out as a promising avenue in applications where the accuracy of the identified linear model is of particular interest. This method employs a formal optimization approach, enabling not only accurate model estimation but also precise estimate of model parameter uncertainties. Recognizing that the estimation accuracy depends on the excitation profile during the identification experiment, this framework can be expanded to design the optimal system excitation, accounting for practical constraints. In this presentation, we will delve into the capabilities of Prediction Error Methods in terms of estimating model accuracy and showcase several applications where this methodology has proven useful.
- Curriculum Vitae: Vedran S. Perić received the master's degree from the University of Novi Sad, Serbia, and the Ph.D. degree from the KTH Royal Institute of Technology, Stockholm in 2016. He was a Research and Teaching Assistant with the University of Novi Sad and Visiting Researcher with the Delft University of Technology. He held positions of Senior Power System Engineer with GE Grid Solutions Research and Development Department, Senior Power System Consultant at GE Energy Consulting, and a position of Senior Business Analyst with Regional Security Coordinator, TSCNET Services GmbH. He was a Head of Research Center for Combined Smart Energy Systems (CoSES) at the TUM Institute of Integrated Materials, Energy and Process Engineering (MEP). He is currently an Associate Professor at the Aarhus University in Denmark. His research interests include a wide range of topics related to power systems dynamic stability, operation and control of smart grids, with the particular focus on integration of different energy systems.
2023/08/15: Prof. Dr. Pablo Zometa, German International University Berlin
- 2023/08/15: Prof. Dr. Pablo Zometa (German International University Berlin) will give a talk on Model-predictive path-following control on a microcontroller using quantized deep neural networks at 1:30 pm in 2/W059.
- Abstract: Model Predictive Path-Following Control (MPFC) is an advanced method for motion control of robots. However, implementing MPFC in real-time on small-scale autonomous platforms with inexpensive embedded hardware remains difficult, despite significant advancements in tailored numerical methods for predictive control. In this work, we employ deep learning techniques to implement predictive path-following control on microcontrollers. We demonstrate that quantized deep neural networks can accurately approximate the MPFC feedback law. Additionally, we address the challenges that arise when the target platform utilizes limited precision arithmetic. Specifically, we utilize post-stabilization with an additional feedback law to mitigate undesired quantization effects. Simulation examples validate the effectiveness of the proposed approach.
- Curriculum Vitae: Pablo Zometa holds an M.Sc. in mechatronic systems from the University of Siegen, Germany (2007), and a Ph.D. from the Otto-von-Guericke University Magdeburg, Germany (2017). His doctoral research focused on automatic code generation for model predictive control of embedded systems. In 2017, he co-founded an IoT startup, assuming a leadership role in technical development. Since 2021, he has served as an associate professor of robotics and mechatronics at the German International University in Berlin. His primary research interests encompass optimization-based predictive control and estimation for embedded mechatronic systems.
2023/06/28: Dr.-Ing Felix Krujatz, Technische Universität Dresden
- 2023/06/28: Dr.-Ing. Felix Krujatz (Technische Universität Dresden) will give a talk on (Mikro-)Algen - vom nachwachsenden Rohstoff zur zellulären Fabrik at 1:00 pm in 2/W040.
- Abstract: Die Rohstoffversorgung rückt im Zuge der aktuellen globalen Entwicklungen immer mehr in den Fokus von Wirtschaft, Politik und Forschung. Durch die fortschreitenden Auswirkungen des Klimawandels in Deutschland sowie die unsichere Versorgungslage aus Drittstaaten mit natürlichen Rohstoffen ist es für die Umsetzung einer nachhaltig wirtschaftenden Bioökonomie in Deutschland unausweichlich die regionale Rohstoffbasis zu erweitern. (Mikro-) Algen gelten nach Zucker- und Lignocellulose-Quellen als 3./4. Generation natürlicher Rohstoffe mit einem breitem industriellen Einsatzpotenzial. Die Produktion der aquatischen Biomasse erfolgt entkoppelt von Agrarfläche in technischen Systemen, sog. Photobioreaktoren (PBR), auf Basis photosynthetisch aktiver Strahlung als Energie- sowie CO2 als Kohlenstoffquelle. Dabei werden offene und geschlossene PBR-Konzepte unterschieden, deren Vor- und Nachteile beleuchtet werden. Die 4. Generation natürlicher Ressourcen setzt nicht auf die Bildung aquatischer Biomasse, sondern auf die photobiokatalytische Wirkung von Cyanobakterien und Mikroalgen zur direkten Konversion von CO2 in Plattformmoleküle. Der Vortrag soll eine übersicht über die Konzepte zur 3. und 4. Generation an natürlichen Rohstoffen geben und das Potenzial von photosynthetisch aktiven Mikroorganismen anhand einiger Beispielprozesse verdeutlichen. Abschließend werden aktuelle Forschungsprojekte der TU Dresden, v.a. im Kontext zur Digitalisierung der Bioprozesse und PBR-Anlagen, vorgestellt.
- Curriculum Vitae: Dr. Felix Krujatz studierte Bio Process Engineering an der Technischen Universität Braunschweig, Deutschland. Im Jahr 2011 machte er seinen M. Sc. und wechselte für seine Promotion mit dem Thema "Development and validation of new bioreactor concepts for phototrophic microorganisms" an die Technische Universität Dresden zu Prof. Thomas Bley. Er schloss seine Promotion 2016 ab und arbeitet seitdem als Forscher und Gruppenleiter in der Algenbiotechnologie am Institut für Natural Materials Technology der TU Dresden. 2021 wurde er zudem Geschäftsführer und Mitbegründer der biotope gGmbH - Center of Applied Aquaculture & Bioeconomy. Darüber hinaus hat er von 2022 bis August 2023 eine Vertretungsprofessur - Umweltbioverfahrenstechnik - an der Hochschule Zittau/Görlitz inne.
2023/06/19: Prof. Dr. Jaime Moreno, Universidad Nacional Autónoma de México
- 2023/06/19: Prof. Dr. Jaime Moreno (Universidad Nacional Autónoma de México) will give a talk on On multivalued observers for nonlinear systems at 2:00 pm in 2/W014.
- Abstract: In this talk we consider a class of not observable nonlinear systems, having a finite number of (non-converging) indistinguishable trajectories. For these systems it is not possible to construct a (global) single-valued observer. We propose, instead, a multivalued observer, allowing to estimate exactly the full set of indistinguishable trajectories. By extending the system’s dynamics, it is also possible to estimate all possible unknown inputs acting on the system. We motivate the problem using real systems, and illustrate the results using simulations and experimental results.
- Curriculum Vitae: Jaime A. Moreno was born in Colombia and he received his PhD degree (Summa cum Laude) in Electrical Engineering (Automatic Control) from the Helmut-Schmidt University in Hamburg, Germany in 1995. The Diplom-Degree in Electrical Engineering (Automatic Control) from the Universität zu Karlsruhe, Karlsruhe, Germany in 1990, and the Licentiate-Degree (with honors) in Electronic Engineering from the Universidad Pontificia Bolivariana, Medellin, Colombia in 1987. He is full Professor of Automatic Control in the Electrical and Computing Department at the Institute of Engineering from the National University of Mexico (UNAM), in Mexico City. He is a member of IFAC Council and the author and editor of 8 books, 12 book chapters, 1 patent, and author and co-author of more than 450 papers in refereed journals and conference proceedings. His current research interests include robust and non-linear control, in particular, with emphasis on Lyapunov methods for higher order sliding modes control, with applications to biochemical (wastewater treatment processes) and electromechanical processes, and the design of nonlinear observers.
2023/06/07: M.Sc. Ferdinand Rewicki, DLR Institute of Data Science
- 2023/06/07: M.Sc. Ferdinand Rewicki (DLR Institute of Data Science) will give a talk on Machine Learning for Space Gardening at 01:00 pm in 2/W040.
- Abstract: Sustained human presence in space requires the development of new technologies to maintain environment control, provide water, oxygen, food and to keep astronauts healthy and psychologically fit. The EDEN NEXT GEN project works along the roadmap of building a flight-ready bio-regenerative life support system within this decade. Being part of that project, we are concerned with detecting unhealthy system states and plant stress in the context of extraterrestrial horticulture.
In this talk, Mr. Rewicki will introduce different classical and deep-learning-based methods for finding anomalies in time series and presents the latest results on differences regarding the types of anomalies these methods can find.
- Curriculum Vitae: Ferdinand Rewicki studied Computational and Data Science at Friedrich Schiller University in Jena, Germany. In 2021, he obtained an M. Sc. with the focus on Machine Learning. From 2017 until 2020, he worked as a software developer in the in the Treasury and Finance domain. Since September 2021, he has been a PhD student in the Institute of Data Science in Jena of the German Aerospace Center. He already worked as a Research assistance on uncertainty estimation and quantification methods for deep neural networks since January 2021.
2023/04/27: M.Sc. Dario Dennstädt, Universität Paderborn & Technische Universität Ilmenau
- 2023/04/27: M.Sc. Dario Dennstädt (Universität Paderborn & Technische Universität Ilmenau) will give a talk on Funnel-MPC at 02:15 pm in 2/NK003.
- Abstract: Model Predictive Control (MPC) is nowadays a widely used control technique for linear and nonlinear systems due to its ability to handle multi-input multi-output systems under control and state constraints. Given a model of a system, the idea is to predict the future system behavior and iteratively solve Optimal Control Problems (OCPs). To use an MPC algorithm initial and recursive feasibility have to be ensured and it is inherently necessary to have a sufficiently accurate model in order to compute an optimal control signal. Complementary, funnel control is a model-free adaptive output-error feedback controller. Under certain structural assumptions this concept guarantees the tracking of a prescribed reference signal within predefined bounds. In this talk the novel concept Funnel-MPC (FMPC) is presented, which combines both approaches, i.e. Model Predictive Control and funnel control. Utilizing a stage cost design in the OPC which mimics the high-gain idea of funnel control, this new approach achieves guaranteed output tracking of smooth reference signals with prescribed performance. This is achieved without imposing any terminal conditions, requirements on the length of the prediction horizon, or utilization of output constraints in the OCP. Most recent advances on Robust-FMPC allow for the application of Funnel MPC in the presence of unknown disturbances and even a structural plant-model mismatch. Current research aims to additionally incorporate learning techniques to allow for steady improvement of the used model and, thus, controller performance while guaranteeing output tracking within predetermined boundaries.
- Curriculum Vitae: Dario Dennstädt studied Mathematics with a minor in Computer Science at Technische Universität Ilmenau, Germany. In 2016, he obtained an M. Sc. with the focus on the algebraic theory of linear time-variant systems. From 2016 until 2020, he worked as a software developer in the field of multilingual XML content and data management at Acolada GmbH, Nürnberg, Germany. Since November 2020, he has been a PhD student in the research group Optimization-based Control of Prof. Dr. Karl Worthmann at the Technische Universität Ilmenau, Germany. Trying to combine model predictive control with the adaptive control technique funnel control in the new concept Funnel-MPC, he has also working as a research assistant in the group of Jun.–Prof. Dr. Thomas Berger at Universität Paderborn, Germany since November 2022.
2023/04/26: Dr.-Ing. Daniel Schubert, DLR Institute of Space Systems
- 2023/04/26: Dr.-Ing. Daniel Schubert (DLR Institute of Space Systems) will give a talk on From the South Pole into Space – Vegetable Cultivation in Antarctica at 11:15 am in 2/N013 (in conjunction with Physikalisches Kolloquium).
- Abstract: A greenhouse in Antarctica provides insights for future plant cultivation on the moon and Mars. Heavy snowstorms with wind speeds of up to 150 km/h, temperatures below -40 °C, total isolation cut off from civilization, including 10 weeks in complete darkness and no rescue capsule like on the International Space Station (ISS), with which a quick return home is possible. These are just some of the challenges faced by the EDEN ISS team during the space test mission at the German Neumayer Station III of the Alfred Wegner Institute in Antarctica. The research project, funded by the EU and led by the German Aerospace Center (DLR), is investigating how plants can be grown in future habitats on the moon and Mars. The longer-term presence of humans in space requires the development of new regenerative technologies to support food and oxygen production, waste recycling, carbon dioxide reduction, waste management, water purification and to keep the crew healthy and psychologically fit. EDEN ISS partners developed for the project the following components: Novel nutrient supply, high-power LED lighting system, biodetection and decontamination, imaging systems to monitor healthy plant growth. During the test mission in Antarctica, the greenhouse system provided various fresh vegetables for the 10 members of the hibernating crew. Project leader Dr. Daniel Schubert presents a general overview of the project in his talk.
- Curriculum Vitae: Dr.-Ing. Daniel Schubert studied at the Technical University of Berlin and has an engineering diploma in industrial engineering with an emphasis on aerospace and production techniques. In 2011, he initiated the EDEN group at the DLR Institute of Space Systems for research on Bio-regenerative Life Support Systems and since served as the team leader of this group. His research expertise is set on habitat interface analysis and plant accommodation and dynamic plant production planning. In the same field he accomplished his PhD at the University of Bremen in 2018. Throughout many projects for ESA, EU, Bundesministerium für Bildung und Forschung, Wirtschaftsförderung Bremen, Dr. Schubert and his team developed many greenhouse concepts, vertical farming feasibility studies, and habitat technologies. Outstanding is the EDEN ISS project. He led this project with 15 international partners, including the organization of the deployment mission of the greenhouse system at the Antarctic research station Neumayer III in 2017/18.
2023/03/22: Dr. Manuel Schaller, Technische Universität Ilmenau
- 2023/03/22: Dr. Manuel Schaller (Technische Universität Ilmenau) will give a talk on Data-based predictive control using eDMD with guarantees at 3:00 pm in 2/W043.
- Abstract: Extended Dynamic Mode Decomposition, embedded in the Koopman framework, is a widely-applied approximation technique to predict the evolution of an observable along the flow of a dynamical (control) system. We provide a complete and rigorous analysis of the approximation error for control-affine systems in which the error is split up into two sources: a projection error coming from a finite dictionary and an approximation error stemming from a finite amount of i.i.d. data used to generate the surrogate model. Using concentration inequalities (e.g.. the Chebyshev inequality) and a finite-element dictionary, we obtain a probabilistic error bound. Finally, we sketch extensions towards data-based dictionaries in Reproducing Kernel Hilbert Spaces (RKHS) and applications in predictive control indicating the usefulness of the derived error bounds to guarantee, e.g., practical asymptotic stability.
- Curriculum Vitae: Manuel Schaller obtained a M.Sc. in Mathematics (with a minor in Computer Science) in 2017 from the University of Bayreuth with focus on PDE-constrained optimization and numerics. In 2021, he received a PhD in Applied Mathematics at the University of Bayreuth under the joint supervision of Prof. Dr. Lars Grüne and Prof. Dr. Anton Schiela. During this time, his research was focused on stability and sensitivity analysis of (infinite-dimensional) optimal control problems, in particular turnpike theory and efficient space-time finite element discretizations for Model Predictive Control. Since July 2020 he is with Prof. Dr. Karl Worthmann at the Technical University of Ilmenau working in the field of Optimization-based control, where since February 2022 he holds a position as Lecturer (akademischer Rat). In his research, he focusses on singular optimal control of port-Hamiltonian systems and guarantees for data-based surrogate models for control systems with particular applications in retinal laser treatment and adaptive high-rise buildings. For his research, he was appointed as GAMM Junior for the years 2020-2023.
2023/03/01: Prof. Dr. Florin Stoican, Politehnica University of Bucharest
- 2023/03/01: Prof. Dr. Florin Stoican (Politehnica University of Bucharest) will give a talk on Computing the explicit MPC solution using the Hasse diagram of the lifted feasible domain at 2:30 pm in 2/W017.
- Abstract: This presentation provides a combinatorial interpretation for the explicit solution of the quadratic cost, linear-constrained MPC (model predictive control) problem. We link the Hasse diagram of the lifted feasible domain with the critical regions which partition the parameter space and serve as polyhedral support for the piecewise affine explicit MPC solution.
- Curriculum Vitae: Florin Stoican received his Electrical Engineering Degree (Systems and Applied Informatics Specialization) from the University "Politehnica" of Bucharest (UPB), Romania in 2008 and his PhD in Control Engineering from Supelec, Gif-sur-Yvette, France in 2011. He continued with a post-doctoral fellowship at NTNU, Norway and obtained his Habilitation thesis in 2018. Currently he is (previously Associate, 2016-2019 and Assistant, 2013-2016) a Full Professor at "Politehnica" University of Bucharest in the Department of Automatic Control and Systems Engineering. His main research interests are the application of set theoretic methods into the fault tolerant control of dynamical systems through the prism of set theoretic elements. His current work involves further research in set theory, constrained optimization problems and the coupling between motion planning, flat representations and spline approximations.
2023/02/27: Prof. Dr. Florin Stoican, Politehnica University of Bucharest
- 2023/02/27: Prof. Dr. Florin Stoican (Politehnica University of Bucharest) will give a talk on A mixed-integer MPC with polyhedral potential field cost for collision avoidance at 2:00 pm in 2/W059.
- Abstract: This work pertains to the use of polyhedral potential fields which take into account the obstacles' shape through their associated sum function. As contributions, we extend the polyhedral sum function notion to the multi-obstacle case and show that the polyhedral support of its piecewise affine representation comes from an associated hyperplane arrangement. We exploit the obtained combinatorial structure to provide equivalent mixed-integer representations and model a repulsive potential term for subsequent use in a Model Predictive Control (MPC) formulation. The advantages are related to a decrease in the computational complexity and a coherent description of the repulsive potential, regardless of the cluttered environment's complexity.
- Curriculum Vitae: Florin Stoican received his Electrical Engineering Degree (Systems and Applied Informatics Specialization) from the University "Politehnica" of Bucharest (UPB), Romania in 2008 and his PhD in Control Engineering from Supelec, Gif-sur-Yvette, France in 2011. He continued with a post-doctoral fellowship at NTNU, Norway and obtained his Habilitation thesis in 2018. Currently he is (previously Associate, 2016-2019 and Assistant, 2013-2016) a Full Professor at "Politehnica" University of Bucharest in the Department of Automatic Control and Systems Engineering. His main research interests are the application of set theoretic methods into the fault tolerant control of dynamical systems through the prism of set theoretic elements. His current work involves further research in set theory, constrained optimization problems and the coupling between motion planning, flat representations and spline approximations.
2023/02/08: Prof. Dr.-Ing. Timm Faulwasser, Technische Universität Dortmund
- 2023/02/08: Prof. Dr.-Ing. Timm Faulwasser (Technische Universität Dortmund) will give a mini workshop on Economic Operation and Hierarchical Control - A Primer from 9:00 am to 12:30 am in 2/W043.
- Abstract: The steadily growing need to operate complex systems close to safety critical constraints while optimizing economic performance is driving manifold research efforts in systems and control, process systems engineering, computer science, and other disciplines. This mini-workshop takes the classic perspective of hierarchical optimization-based control of process systems to provide an overview and introduction. In the first part we recall the automation pyramid and we discuss the practical issues of uncertainty in model-based performance optimization. We then turn towards real-time optimization (or measurement-based optimization) and in particular we focus on the concept of modifier adaptation. The part concludes with results from industrial case studies. The second part shifts the focus towards dissipativity-based analysis of optimal control problems. Specifically, we recall the concept of turnpike properties of optimal control problems which refers to the phenomenon that for varying initial conditions and horizon lengths different optimal solutions converge to the same steady state. While first observations of this phenomenon can be traced back to the 1940s and works of von Neumann and Ramsey, recent progress has been achieved using dissipativity concepts. We also comment on the link between dissipativity and stability in infinite-horizon optimal control. Finally, we close the loop, i.e., we turn towards the dissipatvity- and turnpike-based analysis of economic MPC schemes. The third part links the above viewpoints, i.e., we show how to include ideas from modifier adaptation into economic MPC formulations. The mini-workshop concludes with an outlook on open problems and a feedback discussion.
- Curriculum Vitae: Timm Faulwasser has studied Engineering Cybernetics at the University of Stuttgart, with majors in systems and control and philosophy. From 2008 until 2012 he was a member of the International Max Planck Research School for Analysis, Design and Optimization in Chemical and Biochemical Process Engineering Magdeburg. In 2012 he obtained his PhD from the Department of Electrical Engineering and Information Engineering, Otto-von-Guericke-University Magdeburg, Germany. From 2013 to 2016 he was with the Laboratoire d'Automatique, École Polytechnique Fédérale de Lausanne (EPFL), Switzerland, while 2015-2019 he was leading the Optimization and Control Group at the Institute for Automation and Applied Informatics at Karlsruhe Institute of Technology (KIT), where he successfully completed his habilitation in the Department of Informatics in 2020. In November 2019 he joined the Department of Electrical Engineering and Information Technology at TU Dortmund University, Germany. Currently, he serves as associate editor for the IEEE Transactions on Automatic Control, for the IEEE Control System Letters, and for Mathematics of Control Systems and Signals. His main research interests are optimization-based and predictive control of nonlinear systems and networks with applications in energy, process systems engineering, mechatronics, and beyond.
2023/02/08: Prof. Dr.-Ing. Timm Faulwasser, Technische Universität Dortmund
- 2023/02/08: Prof. Dr.-Ing. Timm Faulwasser (Technische Universität Dortmund) will give a talk on Output Feedback for Stochastic Systems: Data-Driven Control Through the Eyes of Wiener and Willems at 3:30 pm in 2/W043.
- Abstract: The fundamental lemma proposed by Jan Willems and co-authors in 2005 is pivotal for many recent research efforts on data-driven predictive control. It states that, under suitable assumptions, any input-output trajectory of a linear time-invariant (LTI) system can be described as a linear combination of previously recorded trajectories. Recently, there have been several extensions of the fundamental lemma, e.g., to linear-parameter varying systems and to nonlinear systems. Moreover, predictive control based on such non-parametric system descriptions is receiving substantial research interest. In this seminar, we will discuss progress of data-driven output-feedback MPC of stochastic LTI systems. Leveraging polynomial chaos expansions (PCE) of random variables, the origins of which date back to Norbert Wiener, we answer the question of how to formulate the fundamental lemma for stochastic systems. The seminar will illustrate that the knowledge or estimation of past noise realizations allows the construction of Hankel matrices which in turn enable propagation of non-Gaussian and Gaussian uncertainties with non-parametric system descriptions. The seminar will show how to achieve uncertainty propagation for stochastic LTI systems without explicit model knowledge. The final part of the talk turns towards data-driven stochastic MPC. The findings will be illustrated by examples from different applications.
- Curriculum Vitae: Timm Faulwasser has studied Engineering Cybernetics at the University of Stuttgart, with majors in systems and control and philosophy. From 2008 until 2012 he was a member of the International Max Planck Research School for Analysis, Design and Optimization in Chemical and Biochemical Process Engineering Magdeburg. In 2012 he obtained his PhD from the Department of Electrical Engineering and Information Engineering, Otto-von-Guericke-University Magdeburg, Germany. From 2013 to 2016 he was with the Laboratoire d'Automatique, École Polytechnique Fédérale de Lausanne (EPFL), Switzerland, while 2015-2019 he was leading the Optimization and Control Group at the Institute for Automation and Applied Informatics at Karlsruhe Institute of Technology (KIT), where he successfully completed his habilitation in the Department of Informatics in 2020. In November 2019 he joined the Department of Electrical Engineering and Information Technology at TU Dortmund University, Germany. Currently, he serves as associate editor for the IEEE Transactions on Automatic Control, for the IEEE Control System Letters, and for Mathematics of Control Systems and Signals. His main research interests are optimization-based and predictive control of nonlinear systems and networks with applications in energy, process systems engineering, mechatronics, and beyond.
2022/12/16: Prof. Dr. Matthias Gerdts, Universität der Bundeswehr München
- 2022/12/16: Prof. Dr. Matthias Gerdts (Universität der Bundeswehr München) will give a talk on Coordination of interacting systems using optimal control techniques at 2:00 pm in 2/N001.
- Abstract: Automation and autonomy becomes more and more important in many robotics applications, especially for mobile robots, which move automatically in a production site, for automated cars, which move in the traffic, or in space robotics. These mobile robots often do not just follow a precomputed reference path, but they need to be able to update their trajectories in order to react on changing environments. This typically requires a feedback control strategy, which takes into account the current state of the robot and the environment. To this end we employ a model-predictive control (MPC) strategy, which requires to solve (discretized) optimal control problems repeatedly. A particular focus of the talk is on methods for the coordination of interacting systems, which are not necessarily cooperative. We investigate suitable solution concepts and embed them into the MPC framework. The first approach uses generalized Nash equilibrium problems, which allow to model the coordination of automated agents without using pre-defined priorities. The second approach couples scheduling tasks with optimal control and leads to a bi-level optimization formulation. Numerical experiments and case studies will be presented to illustrate the methods.
- Curriculum Vitae: Prof. Dr. Matthias Gerdts studied Mathematics with minor Computer Science at the University of Technology Clausthal, Germany, and graduated in 1997. He received his doctoral degree in 2001 and his Habilitation in 2006 from the University of Bayreuth, Germany. In 2003 he was a visiting professor at the University of California, San Diego. From 2004 to 2007 he held a junior professorship for numerical optimal control at the Department of Mathematics of the University of Hamburg, Germany, and moved to a lecturer position for mathematical optimization at the University of Birmingham, U.K., from 2007 to 2009. From 2009 to 2010 he was an associate professor for optimal control at the University of Würzburg, Germany. Since 2010 he is a full professor for engineering mathematics at the Department of Aerospace Engineering of the Bundeswehr University Munich, Germany. His primary research interests are optimal control, optimization techniques, model-predictive control, differential-algebraic equations, and sensitivity analysis with applications in automotive systems, robotics, and aerospace engineering.
2022/11/29: Jun.-Prof. Dr. Tsu-Wei Chen, HU zu Berlin
- 2022/11/29: Prof. Dr. Tsu-Wei Chen (Humbold Universität zu Berlin) will give a talk on Multi-scale modelling and phenotyping facilitate understanding of plant-plant interactions and canopy productivity at 2:00 pm in 2/W040.
- Abstract: Recent advances in high-throughput plant phenotyping and computational power for modelling activities provide new avenues for studying crop physiology. In this work, we demonstrate how to combine phenotyping platform, multi-scale and multi-dimensional modelling approaches, scientific workflow systems to study the inter- and intra-genotypic interactions of maize, wheat and cucumber grown under homogeneous and heterogeneous canopy. Furthermore, we show how to use the new insights into plant-plant interactions facilitate designing cropping systems and plant breeding.
- Curriculum Vitae: Tsu-Wei Chen holds the renowned DFG Emmy Noether Programm and is professor of Intensive Plant Food Systems of Faculty of Life Science at Humboldt Universität zu Berlin since September 2020. His research group works on up-scaling the environmental effects on leaf photosynthesis to canopy level using modelling approaches and systems analyses. His interdisciplinary collaboration partners include computer scientists, mathematicians and crop geneticists. Besides classical modelling approaches, he is specialized in functional-structural plant models, which he uses to analyze the triple feedbacks between environmental triggers, dynamic canopy architectures and physiological functions. His research focusses on model-assisted high-throughput phenotyping and physiological mechanisms optimizing canopy photosynthesis through acclimation strategies.
2022/11/14: Jun.-Prof. Dr. Thomas Berger, Universität Paderborn
- 2022/11/14: Jun.-Prof. Dr. Thomas Berger (Universität Paderborn) will give a talk on Funnel control and applications at 5:00 pm in 2/W040.
- Abstract: The control of dynamical systems with prescribed performance requirements is an active research area of systems and control theory. In view of necessary safety guarantees, suitable control techniques are of high practical relevance. Concurrently, a certain robustness is necessary because a precise measurement of the full state is often not available and many system parameters are unknown or uncertain. In this talk, the method of funnel control will be presented as a control technique which exhibits these requisite properties and only requires the knowledge of some structural invariants of the system class. This class can contain both finite and infinite dimensional systems, as the dimension of the state does not need to be known. For infinite dimensional systems we distinguish between systems which have a well defined relative degree (and are hence amenable to funnel control by straightforward arguments) and systems for which this is not the case - and hence individual methods are required.
- Curriculum Vitae: Thomas Berger was born in Germany in 1986. He received his B.Sc. (2008), M.Sc. (2010), and Ph.D. (2013), all in Mathematics and from Technische Universität Ilmenau, Germany. From 2013 to 2018, Dr. Berger was a postdoctoral researcher at the Department of Mathematics, Universität Hamburg, Germany. Since January 2019 he is a Juniorprofessor at the Institute for Mathematics, Universität Paderborn, Germany. His research interest encompasses systems and control theory, differential–algebraic systems and multibody dynamics. For his exceptional scientific achievements in the field of Applied Mathematics and Mechanics, Dr. Berger received the "Richard-von-Mises Prize 2021" of the International Association of Applied Mathematics and Mechanics (GAMM). He further received several awards for his dissertation, including the "2015 European Ph.D. Award on Control for Complex and Heterogeneous Systems" from the European Embedded Control Institute and the "Dr.-Körper-Preis 2015" from the GAMM. He serves as an Associate Editor for Mathematics of Control, Signals, and Systems and the IMA Journal of Mathematical Control and Information, and as a Review Editor for Frontiers in Control Engineering.
2022/11/10: Dr. Yuning Jiang, EPFL (Lausanne)
- 2022/11/10: Dr. Yuning Jiang (École Polytechnique Fédérale de Lausanne) will give a talk on Data-driven control and optimization with kernel-based learning at 2:30 pm in 2/N111.
- Abstract: Kernel-based learning is a nonparametric learning technique widely used to interpolate an unknown nonlinear function based on a finite number of samples. This talk will first introduce the deterministic error bounds of the kernel models. These bounds could provide quantification of the state-dependent uncertainty for an unknown nonlinear dynamic system. Based on this advantage, the second part of this talk will present a data-driven predictive controller using kernel-based system identification that enjoys deterministic guarantees of safety. And based on these bounds, the last part of this talk will introduce a black-box global optimization approach in RKHS with an explicit sublinear bound of the cumulative regret.
- Curriculum Vitae: Yuning Jiang received his B.Sc. degree in electrical engineering from Shandong University, China, in 2014 and his Ph.D. degree in information engineering from the University of Chinese Academy of Sciences in 2020. During his Ph.D. study, he was a Visiting Scholar with UC Berkeley, the University of Freiburg, and TU Ilmenau. He is currently a Postdoctoral Researcher with the Automatic Control Lab, EPFL, Switzerland. His research focuses on optimization- and learning-based control and operation for complex systems, including nonlinear autonomous systems (e.g., autonomous vehicles, robotics, and smart buildings) and large-scale multi-agent systems (e.g., power and energy systems, IoT, and traffic networks).
2020/02/27: Prof. Dr. Florin Stoican, Politehnica University of Bucharest
- 2020/02/27: Prof. Dr. Florin Stoican (Politehnica University of Bucharest) will give a talk on About the use of B-spline functions in motion planning at 4:00 pm in 2/W065.
- Abstract: B-spline functions are a popular choice for describing trajectories associated with nonlinear dynamics and for imposing continuous-time constraint validation. Exploiting their properties (local support, convexity and positivity) usually leads to sufficient conditions which are conservative. In this talk I will present two methods which reduce the conservatism: the first makes use of sum-of-squares polynomials to provide a linear matrix inequality-type formulation giving necessary and sufficient conditions and the second makes use of knot refinement strategies to arbitrarily improve on the sufficient conditions. In both cases, the approaches are validated for a motion planning problem where off-line trajectories with obstacle(s) avoidance guarantees are generated.
- Curriculum Vitae: Florin Stoican received his B.S. degree from the Univ. "Politehnica" of Bucharest (UPB), Romania in 2008 and his PhD in Control Engineering from Supelec, Gif-sur-Yvette, France in 2011. He continued with a post-doctoral fellowship at NTNU, Norway. From 2013 he is part of the Automatic Control and Systems Engineering department of (UPB). His research interests are in the fields of fault tolerant control, set theoretic methods and, more recently, mixed integer programming and motion planning problems. He is the (co-)author of 2 books, 18 journal papers and over 50 conference papers.
2020/01/30: Prof. Dr.-Ing. Johann Reger, TU Ilmenau
- 2020/01/30: Prof. Dr.-Ing. Johann Reger (TU Ilmenau) will give a talk on Exact Backstepping Control for Systems in Pure Feedback Form at 5:30 pm in 2/W014.
- Abstract: Traditional backstepping approaches may struggle to asymptotically stabilize systems in pure feedback form, due to its inherent implicit equations. Approximation based designs only have a limited domain of validity and turn out sensitive to model uncertainty and disturbances. We propose a new design that circumvents the necessity of solving implicit algebraic equations by introducing new state variables. Additional augmentations to the backstepping Lyapunov design lead to explicit expressions for the associated differential equations. The result is a dynamic state feedback, capable of asymptotically stabilizing the origin of a general class of nonlinear systems, based on just standard assumptions.
- Curriculum Vitae: Dr. Johann Reger received his diploma degree (Dipl.-Ing.) in Mechanical Engineering in 1999 and his doctorate (Dr.-Ing.) in Control Engineering in 2004, both from the University of Erlangen-Nuremberg in Germany. He has held several postdoc positions, among others, with the Mechatronics Department at CINVESTAV-IPN in Mexico-City, the EECS Control Laboratory at the University of Michigan in Ann Arbor, and the Control Systems Group at TU Berlin. Since 2008 he is a full professor and head of the Control Engineering Group at the Computer Science and Automation Faculty, TU Ilmenau, in Germany. There he also serves as vice-dean and director of the Institute for Automation and Systems Engineering. His current research foci are on adaptive and robust control, variable structure and sliding mode control, state and parameter estimation. Application areas include robotics, mechatronics, automotive, and water systems.
2019/10/21: Dr. Victor Magron, CNRS-LAAS, Équipe MAC
- 2019/10/21: Dr. Victor Magron (CNRS-LAAS, Équipe MAC) will give a talk on Semidefinite programming and moment/sums-of-squares hierarchy for solving polynomial optimization problems at 3:00 pm in 2/W040.
- Abstract: Polynomial optimization is a challenging and important problem, which consists of computing the infimum of a polynomial function under algebraic constraints. The emergence of this exciting new field goes back to the last decade and has led to striking developments from a cross fertilization between (real) algebraic geometry, applied mathematics, theoretical computer science and engineering. The backbone of this powerful methodology is the "moment-SOS" approach, also known as "Lasserre's hierarchy".
- Curriculum Vitae: Victor Magron received his graduated degree from École Centrale Paris Engineering School, in 2010, while receiving the MSc from Tokyo University (double diploma). In 2013, he received the PhD degree in computer science from École Polytechnique. In 2014, he was a postdoc fellow in CNRS-LAAS, Toulouse, then a research associate at Imperial College London. In 2015, he was appointed junior researcher in CNRS-VERIMAG. Since 2019, he is affiliated to CNRS-LAAS in the MAC team.
2019/07/03: Prof. Dr.-Ing. Peter Protzel, TU Chemnitz
- 2019/07/03: Prof. Dr.-Ing. Peter Protzel (TU Chemnitz) will give a talk on Sind wir bald da? Autonomes Fahren, Maschinelles Lernen und KI - (Ernüchternde) Einblicke zum Stand der Forschung at 3:30 pm in 2/W040.
- Abstract: Autonomes Fahren und KI haben in letzter Zeit einen "Hype-Gipfel" erreicht, der unter Forschern auf dem Gebiet eher kritisch betrachtet wird, weil dadurch oft überzogene Erwartungen und Ängste geweckt werden. Daher soll dieser Vortrag in verständlicher Form anhand von Beispielen folgende Fragen erörtern: Was verstehen wir überhaupt unter KI, welche Art von "Intelligenz" steckt dahinter? Wie sind die jüngsten Erfolge von "Deep Learning" einzuschätzen und wo liegen die Grenzen? Ist ein weiterer Fortschritt nur eine Frage der Zeit (wie bei der Steigerung der Rechenleistung) oder welche fundamentalen Probleme gibt es? Für die meisten Menschen ist es schwieriger einen Schach- oder Go-Weltmeister zu schlagen als einen Führerschein zu machen. Wenn autonomes Fahren "KI auf Rädern" ist, warum kann dann ein Computer, der gerade den Weltmeister in Go geschlagen hat, nicht vernünftig Autofahren? Der Grund ist der fundamentale Unterschied zwischen maschineller und biologischer Informationsverarbeitung. Die genaue Funktionsweise biologischer Gehirne ist nach wie vor ein Rätsel und kann daher auch nicht "nachgebaut" werden. Der KI fehlen grundlegende Fähigkeiten wie die Repräsentation und Adaption von komplexen Modellen unserer Umwelt als Voraussetzung für das Verstehen kausaler Zusammenhänge und für die Planung komplexer Aktionen in offenen, unstrukturierten und nicht-deterministischen Umgebungen. Wie weit ist es also noch vom gegenwärtigen Stand bis zu einer starken, "richtigen" KI? Nach einem schönen Vergleich von Florian Gallwitz in einem kürzlichen Wired Artikel sind wir derzeit davon so weit entfernt wie eine Silvester-Rakete von der interstellaren Raumfahrt.
- Curriculum Vitae: Peter Protzel hat in Bochum und Braunschweig Elektrotechnik studiert und 1987 an der TU Braunschweig promoviert. Anschließend war er fünf Jahre am NASA Langley Research Center in Virginia und danach sieben Jahre am Bayerischen Forschungszentrum für Wissensbasierte Systeme in Erlangen. Seit 1998 ist er Inhaber der Professur für Prozessautomatisierung an der TU Chemnitz.
2019/06/12: Dr.-Ing. Sören Weinrich, Deutsches Biomasseforschungszentrum
- 2019/06/12: Dr.-Ing. Sören Weinrich (Deutsches Biomasseforschungszentrum) will give a talk on Model-based control for demand-oriented power supply of anaerobic digestion plants at 3:30 pm in 2/W185.
- Abstract: The changing conditions within the energy sector in Germany force biogas plants to meet new requirements. Demand-oriented power supply to compensate the divergence between electricity production and consumption by uncontrolled sources like wind and solar power defines one option for future plant operation. Dynamic models of the anaerobic digestion process are essential tools to predict the progression of various process variables and to guarantee secure and optimized plant operation under highly dynamic or demand-oriented feeding. However, due to the high number of state variables and unknown model parameters complex models such as the established Anaerobic Digestion Model No.1 (ADM1) still cannot be applied in practice. Thus, the development of simplified model structures is of great importance for a standardized application of model-based monitoring and control systems at anaerobic digestion plants. The presentation will focus on the systematic simplification and comparison of available model structures of the anaerobic digestion process. Based on the derived model structures a model predictive control (MPC) concept was developed to provide optimal feeding strategies and enable demand-oriented power supply of anaerobic digestion plants. Full-scale application showed high intraday flexibility ranging as far as 30 to 130% of the average gas production rate and high process stability during demand-oriented feeding. The gas storage demand could be reduced significantly by up to 45% in comparison to steady state operation.
- Curriculum Vitae: Sören Weinrich studied agriculture and environmental science at the professorship of 'Abfall- und Stoffstromwirtschaft' at the university of Rostock and received his PhD in 2017. He has since been working at the DBFZ in Leipzig. His research topics focus around modeling and control of biogas plants.
2019/05/08: Prof. Dr.-Ing. Dr. h. c. Oliver Sawodny, University of Stuttgart
- 2019/05/08: Prof. Dr.-Ing. Dr. h. c. Oliver Sawodny (University of Stuttgart) will give a talk on System Dynamic Methods for Optimization in Electric Vehicles at 3:30 pm in 2/W020.
- Abstract: The powertrain of an all-electric vehicle has changed radically compared to a conventional combustion vehicle. With this change new challenges but also new possibilities arise in terms of system integration and system control. Even though battery technologies continuously improve in capacity and price, one of the main challenges and customer concerns is the overall range. Consequently, founding is mostly directed to research in the field of battery capacity and battery management. However, several essential questions seek to improve the all-electric range by means of advanced driver assistant systems. Research activities at the Institute for System Dynamics thus focus on operating as well as energy management strategies for electric powertrain architectures based on software solutions and the consideration of additional environmental data. Latter is obtained from a constantly increasing number of vehicle sensors and increasing vehicle connectivity and is used for example in loss minimizing control strategies of electric drive modules or in cycle and driver specific range predictions. In this context, estimating and optimizing battery lifetime is also essential for energy management. In most automotive applications, the battery can be represented by an electrical equivalent circuit. Lifetime effects are covered by adding empirical aging relations. Li-ion cell measurements are therefore essential and aging tests are exemplarily described for a LiMn2O4 cell. Based on the derived models, an optimization framework for battery lifetime extension is described for a given scenario. As last aspect the thermal management in EV's is discussed. Therefore, model-based approaches to describe the thermal circuits are introduced and methods to optimize the operational strategy and the design presented.
- Curriculum Vitae: Professor Sawodny received his Dipl.-Ing. degree in electrical engineering from the University of Karlsruhe, Karlsruhe, Germany, in 1991 and his Ph.D. degree from the University of Ulm, Ulm, Germany, in 1996. In 2002, he became a Full Professor at the Technical University of Ilmenau, Ilmenau, Germany. Since 2005, he has been the Director of the Institute for System Dynamics, University of Stuttgart, Stuttgart, Germany. His current research interests include methods of differential geometry, trajectory generation, and applications to mechatronic systems. He received important paper awards in major control application journals such as Control Engineering Practice Paper Prize (IFAC, 2005) and IEEE Transaction on Control System Technology Outstanding Paper Award (2013). He is a senior member of IEEE and senior editor of Mechatronics.
2019/04/12: Prof. Jean Lévine, MINES ParisTech
- 2019/04/12: Prof. Jean Lévine (MINES ParisTech) will give a talk on An introduction to flatness with emphasis on computational aspects at 3:30 pm in 2/W017.
- Abstract: The notion of differentially flat (or shortly flat) system concerns a particular class of nonlinear systems often encountered in practice. It has been introduced in the early 90's by Michel Fliess, Philippe Martin, Pierre Rouchon and the author, and has shown to be most useful in motion planning and tracking of nonlinear systems. In this presentation, we recall the basics of this theory and show, as an example, that aircraft dynamics are flat, giving rise to an autopilot design that is different and much simpler than the traditional AFCS's. We also present various applications, in particular to mechatronics and robotics. We conclude our presentation by some remarks on computer algebraic as well as numerical sides of flatness, and on global aspects of this theory versus singularities.
- Curriculum Vitae: Jean Lévine obtained his 'Doctorat d' État' in Mathematics in 1984 for which he was awarded the Best Thesis AFCET Prize, section Theory, in 1985. He has held various professorship positions, has been a Director of Research with MINES-ParisTech, PSL Research University, since 2006 and is in charge of the Systems and Control Doctoral Studies. He is presently an Emeritus Director of Research and joined the 'Fondation Sciences Mathématiques de Paris' (FSMP) at the Institut Henri Poincaré in 2016 as a Math-Industries special adviser. He contributed to many research fields, including Differential Games, Nonlinear filtering and Nonlinear control. He is one of the founders of differential flatness theory and of the notion of barrier in state constrained nonlinear systems. He also contributed to the transfer of advanced control know-how to the industry through many collaborations with French and international companies on applications such as distillation columns, chemical reactors, food and bio-engineering processes, aircraft control, car equipments, crane control, machine tools, magnetic bearings, high-precision positioning systems and web-to-web machines.
2018/12/12: Prof. Dr. Lars Grüne, University of Bayreuth
- 2018/12/12: Prof. Dr. Lars Grüne (University of Bayreuth) will give a talk on Computing nonsmooth control Lyapunov functions on a grid at 1:45 pm in 2/N001.
- Abstract: Control Lyapunov functions play an important role for controller design for nonlinear control systems. They can be interpreted as a road map, which for each point of the state space indicates a set of good directions towards the target from which the value of the controller can be computed.
Lyapunov functions can be represented as sub- or supersolutions of certain partial differential equations (PDEs) and due to this fact various numerical schemes for computing Lyapunov functions have been designed based on discretization schemes for PDEs. However, most of these schemes require appropriate smoothness of the function to be computed. It is known from seminal works of Artstein, Clarke, Sontag and others in the 1980s and 1990s that Control Lyapunov functions are in general not smooth.
The talk explains a new way to cope with this problem. It is based on joint work with Robert Baier, Philipp Braun, and Chris Kellett.
- Curriculum Vitae: Lars Grüne has been Professor for Applied Mathematics at the University of Bayreuth, Germany, since 2002. He received his Diploma and Ph.D. in Mathematics in 1994 and 1996, respectively, from the University of Augsburg and his habilitation from the J.W. Goethe University in Frankfurt/M in 2001. He held visiting positions at the Universities of Rome 'Sapienza' (Italy), Padova (Italy), Melbourne (Australia), Paris IX - Dauphine (France) and Newcastle (Australia). Prof. Grüne is Editor-in-Chief of the journal Mathematics of Control, Signals and Systems (MCSS) and Associate Editor of several other journals, including the Journal of Optimization Theory and Applications (JOTA) and the IEEE Control Systems Letters. His research interests lie in the area of mathematical systems and control theory with a focus on numerical and optimization-based methods for nonlinear systems.
2018/12/05: Prof. Dr.-Ing. habil. Dipl.-Math. Klaus Röbenack, TU Dresden
- 2018/12/05: Prof. Dr.-Ing. habil. Dipl.-Math.Klaus Röbenack (TU Dresden) will give a talk on Lösung regelungstechnischer Probleme mittels Quantorenelimination at 3:30 pm in 2/W021.
- Abstract: Zahlreiche regelungstechnisch relevante Entwurfsaufgaben lassen sich als Entscheidungsprobleme formulieren und mittels Quantorenelimination lösen. Im Vortrag wird eine Auswahl dieser Probleme behandelt und deren Lösung mittels Quantorenelimination vorgestellt.
-Entwurf statischer Ausgangsrückführungen: Für lineare zeitinvariante Zustandsraummodelle sind sowohl Bedingungen für Polplatzierbarkeit bzw. Stabilisierbarkeit mittels Zustandsrückführung als auch die konkrete Berechnung des jeweiligen Regelgesetzes hinlänglich bekannt. Im Fall einer statischen Ausgangsrückführung sind die Existenzbedingungen und die Berechnung wesentlich schwieriger. Sowohl Polplatzierbarkeit als auch Stabilisierbarkeit können als Entscheidungsprobleme formuliert werden.
-Globale Stabilitätsuntersuchungen: Für globale Stabilitätsuntersuchungen bietet sich die zweite Methode von Ljapunov an. Die für den Stabilitätsnachweis benötigten Bedingungen hinsichtlich der Definitheit lassen sich ebenfalls als Entscheidungsprobleme formulieren.
Bei den genannten Problemen treten Aussagen auf, die Quantoren enthalten. Für den Entwurf bzw. die Auslegung einer Regelungseinrichtung sind äquivalente Formulierungen der jeweiligen Stabilitätsbedingungen ohne Quantoren erforderlich. Dieser Übergang von einer Aussage mit Quantoren zu einer äquivalenten quantorenfreien Aussage erfolgt mittels Quantorenelimination. Alfred Tarski zeigte, dass eine solche über dem Körper der reellen Zahlen grundsätzlich möglich ist. Zu ihrer praktischen Durchführung wurden in den letzten Jahrzehnten verschiedene algorithmische Verfahren entwickelt bzw. implementiert. Die zunehmend effizienter werdenden Algorithmen ermöglichen die Anwendung von Quantorenelimination auf praktisch und theoretisch interessante Fragestellungen der Regelungstechnik.
- Curriculum Vitae: Klaus Röbenack hat Elektrotechnik und Mathematik an der TU Dresden studiert, an der er im Jahr 1999 auch erfolgreich zum Thema Beitrag zur Analyse von Deskriptorsystemen promovierte. Nach seiner Habilitation an der Fakultät Elektrotechnik und Informationstechnik im Jahr 2005 wurde Klaus Röbenack 2009 zum Professor für Regelungs- und Steuerungstheorie an der TU Dresden berufen. Momentan ist er dort als Institutsdirektor des Instituts für Regelungs- und Steuerungstheorie tätig. Er ist ein Fachgutachter bei zahlreichen wissenschaftlichen Tagungen und Zeitschriften wie Automatica, Journal of Computational and Applied Mathematics, sowie IEEE Transactions on Automatic Control. Seine Forschungsinteressen reichen von der nichtlinearen Regelungstechnik bis zu algorithmischen Methoden in der Regelungstheorie.
2018/11/07: Prof. Dr.-Ing. Christian Ebenbauer, University of Stuttgart
- 2018/11/07: Prof. Dr.-Ing. Christian Ebenbauer (University of Stuttgart) will give a talk on To commute or not to commute? The role of geometric control in optimization algorithms at 2:15 pm in 2/W014.
- Abstract: Optimization algorithms are often used in control to solve decision making problems. Control methods, however, are used much less in optimization, despite the facts that control is dedicated to the analysis and design of dynamical systems and that optimization algorithms are dynamical systems. In recent years, however, the research about employing feedback ideas in optimization algorithm design has gained some momentum. In this talk, we present some recent results in this direction, in which we specifically employ fundamental ideas from geometric control to develop optimization algorithms. In particular, we show how non-commutative effects in vector fields and maps give rise to novel classes of extremum seeking, derivative-free and distributed optimization algorithms.
- Curriculum Vitae: Christian Ebenbauer received his MS (Dipl.-Ing.) in Telematics (Electrical Engineering and Computer Science) from Graz University of Technology, Austria, in 2000 and his PhD (Dr.-Ing.) in Mechanical Engineering from the University of Stuttgart, Germany, in 2005. After having completed his PhD, he was a Postdoctoral Associate and an Erwin Schrödinger Fellow at the Laboratory for Information and Decision Systems, Massachusetts Institute of Technology (MIT), USA. Since April 2009, he is a full professor at the Institute for Systems Theory and Automatic Control, University of Stuttgart, Germany. His research interests lie in the areas of dynamical systems, control theory, optimization and computation. Recent research projects focus on system-theoretic approaches to optimization algorithms, extremum seeking control, MPC and MHE algorithms.