Skip to main content

Control and Process Engineering

Regelungs- und Prozesstechnik
Profile

Profile

Recent engineering problems become more and more complex, nonlinear and high-dimensional. We develop together with students systems to analyze, simulate and control these technological processes. Control algorithms like Model Predictive Control and fresh high-level programming languages like Julia or MATLAB/Simulink are associated with artificial intelligence, embedded computing or process engineering to span a wide range of research. These fields of research are explained in detail below. Our research topics are presented on the website of the research group for Control and Process Engineering.

Scheme of a plate with three heat sources
Schematische Darstellung einer Platte mit Wärmequellen und Sensoren und dem dazugehörigen Temperaturverlauf.
Quelle:
Stephan Scholz

Thermal Process Engineering

Thermal problems play an important role in many technological processes. Many examples like building air conditioning, cooling of engines or server clusters, and optimal tempering for the production of synthetic material unveil the significance of the right heat supply. Our team develops simulations and control algorithms for thermal problems in the case of multiple heat sources. In these scenarios the controller has to decide how much heat shall be supplied and at which position. Common applications can be found in the field of bake processes for the semiconductor industry.

Scientific Machine Learning

Recent machine learning approaches have a huge impact on IT services like search engines and social networks. Though, the research of complex systems bases as well on data-driven modeling. In many cases it is too complicated to describe these complex systems in detail. Therefore, systems with partly known properties are modeled as so called grey box models to approach the real system with the computation on large data sets. For example the driving dynamics of vehicles cannot be described in detail because every component has some certain influence. Instead a grey box model is considered which is improved with data from experiments and test drives. Chemical reactions with a huge amount of particles or reactants state another example. It is very hard to state a model in closed form for these reactions because every component could interact with each other. Thus, these chemical reactions are modeled as grey box models and optimized with data from experiments.

See also: SciML

Embedded Systems

Embedded systems can be found in almost every technological application ranging from consumer electronics to industrial robotics. They are used for measuring and processing data, for example light management in buildings. In many cases, commercial controllers are less flexible and more expensive than single-board computer. Thus, we use single-board computers as controlling units in scientific projects and small scale industrial processes in the areas of process technology, electrical and mechanical engineering to examine their field of application.

Projects and Theses

Our projects and theses cover the research topics: thermal process engineering, scientific machine learning and embedded systems. Interested students can choose either software related topics with a solid theoretical basis or hardware and lab related topics including experiments and their evaluation.

    Software development

    We develop mathematical models and controllers for complex dynamical systems and implement them in modern and common programming languages like Julia, MATLAB/Simulink or C/C++. Projects and theses are offered in the areas of

    Lab related topics

    We work on the transformation from closed source and proprietary hard- and software towards open source solutions based on platforms like Raspberry Pi or Arduino. Our offered projects and theses cover all topics of systems and control engineering as

    • design and construction of hardware,
    • modeling and system identification of the plants and
    • implementation of controllers and user interfaces.

    The targeted experiments in our lab comprises

    • Electrical and mechanical oscillators,
    • Hydraulic models (water tank, heel control) and
    • Inverse Pendulums.
    Seminar

    The Embedded Control seminar shall enable the students to delve into a topic of recent research and development and to present their results to an audience. Available topics cover various branches of control engineering like Nonlinear Control or related fields like Mathematical Optimization, Numerical Analysis or Machine Learning.

     

    Presentations

    Friday, 21.01.2022

    • Linear–quadratic–Gaussian control: Kalman filter as an observer
    • Ensemble Kalman-Filter: Monte Carlo approximation
    • H-infinity control
    • Mathematical optimization and the Brachistochrone problem
    Friday, 28.01.2022
    • Coupled electrical oscillators: RLC circuits
    • Convex Optimization and the Frank-Wolfe algorithm
    • Data-driven simulation and control for a thermal system

     

    Recent topics
    Mathematical Optimization
    1. Basics and the Brachistochrone problem
    2. Convex Optimization and the Frank-Wolfe algorithm
    Robust and nonlinear control
    1. H-infinity control
    2. Lyapunov-based control using Automatic Differentiation
    Applications of the Kalman filter
    1. Linear–quadratic–Gaussian control: Kalman filter as an observer
    2. Nonlinear observers: Extended and Unscented Kalman filters
    3. Ensemble Kalman-Filter: Monte Carlo approximation
    Modeling and control of coupled systems
    1. Coupled mechanical oscillators: spring-mass-damper systems
    2. Coupled electrical oscillators: RLC circuits
    Numerical methods
    1. Software-in-the-Loop: Numerical solver for linear plant model
    2. Data-driven simulation and control for a thermal system

     

    Selection of previous topics
    • Automatic Differentiation
    • Ensemble, Extended and Unscented Kalman filter
    • Noise driven systems: Wiener process
    • Data driven modeling: Time series, autoregressive, moving-average
    • Data-Driven Control: Dynamic Mode Decomposition and Koopman Theory
    • Neural Ordinary Differential Equations
    • Port-Hamiltonian Systems

     

    Courses

    We offer courses in the field of control engineering and embedded systems.

    List of courses

    Microcontroller Winter / Summer Lecture and Lab Moodle course
    Control Engineering Winter / Summer Lecture and Lab Moodle course
    Advanced Control Systems: Digital Control Summer only Lecture and Lab Moodle course
    Embedded Control Winter only Seminar and Lab Moodle course
    Embedded GUI Summer only Lecture and Lab Moodle course
    Tutorial Control Engineering (optional) Winter / Summer   Moodle course

     

    Control Engineering

    This course gives an introduction to linear continuous and digital control systems in the time and frequency domain. It covers the modeling of dynamical systems as ordinary differential equations and transfer function and the design of controllers with heuristic, algebraic and graphic methods.

    Contents

    1. Modeling of dynamical systems in the time and frequency domain
      • Laplace Transform
      • Transfer functions
      • Block diagrams
         
    2. Analysis of open- and closed-loop linear systems
      • Proportional, integral and differential (PID) behavior
      • Bounded-input-bounded-output (BIBO) stability
      • Bode and Nyquist diagram
         
    3. Algebraic, graphic and heuristic controller design methods
      • Routh-Hurwitz criterion
      • Nyquist criterion
      • Root locus analysis
      • Ziegler-Nichols and Naslin method
         
    4. Digital control systems in the frequency domain
      • Z-Transform
      • Zero-Order Hold
      • Bilinear transform (Tustin’s method)
      • Jury stability criterion
      • Dead-beat control

    Advanced Control Systems

    Modern complex dynamical systems are modeled, simulated and controlled in the time domain as presented in this course. In particular, common state-space methods for the analysis and control of linear time-invariant system are discussed and the modeling and control of nonlinear systems are introduced.

    Contents

    1. Modeling of dynamical systems in the state-space
      • Laplace transform
      • Eigenvalues and Eigenvectors
      • Stability of LTI systems
         
    2. Design of feedback systems
      • Controllability
      • Full State feedback
      • Ackermann's formula
      • MIMO systems via Eigenvalue decomposition
         
    3. Design of state observers
      • Observability
      • Luenberger State Observer
      • Separation principle
         
    4. Introduction to optimal control
      • Linear-Quadratic Regulator
         
    5. Nonlinear systems and discretization methods
      • Zero-Order Hold
      • Lyapunov Stability
      • Van-der-Pol Oscillator
         
    6. Introduction to nonlinear control systems
      • Sliding-Mode Control
      • Nonlinear Model Predictive Control

    Contact & People

    General contact details

    Phone
    Room H240
    On campus
    Building H, Main building
    Doggenriedstraße
    88250 Weingarten
    Germany
    Postal address RWU Hochschule Ravensburg-Weingarten
    University of Applied Sciences
    Control and Process Engineering
    P.O. Box 30 22
    88216 Weingarten
    Germany

    Lab Team

    Prof. Dr.-Ing. Lothar Berger

    Control engineering, process engineering, mathematical methods
    Head of the Control and Process Engineering Laboratory
    Lothar Berger

    Stephan Scholz M.Sc.

    Academic Staff Member, PhD Student: Modeling, Numerical Simulation, Control Engineering
    Stephan Scholz