Physical and Data-Based Modeling (PDM)

  • Type: Lecture + Exercise
  • Chair: IRS
  • Semester: ST 24
  • Time/Place:

    Wednesday, 11:30 – 13:00 at Messtechnik, MTI (30.33)
    Thursday,     14:00 – 15:30 at Fritz-Haller-Hörsaal (20.40)

    Additionally live stream via Zoom:
    Meeting-ID: 675 5281 0114

    The first lecture will take place on 17.04.2024.  

  • Start: 17.04.2024
  • Lecturer:

    Prof. Dr.-Ing. Sören Hohmann
    M. Sc. Armin Gießler

  • SWS: 2+1
  • ECTS: 6
  • Lv-No.: 2303166
  • Exam: Oral exam
  • Information:

    This lecture is held in English



Contact  If you have any questions concerning the lecture or the exercise, please contact Armin Gießler

In contrast to the former “Modellbildung und Identifikation”, this course requires a profound knowledge in multivariable systems and optimization. Thus, attendance of the lecture Optimization of Dynamic Systems (ODS) is an absolute precondition to appropriately follow the course! Prior knowledge about (linear) state space representations and realizations, the concept of “zeros” in the state space, and observability is highly recommended (see e.g. Regelung linearer Mehrgrößensysteme (RLM))!

Furthermore, sound understanding of Higher Mathematics I-III, linear electrical network theory and engineering mechanics / physics is required to successfully attend the lecture, exercise tasks / case studies, and exam.

Teaching Content

This course aims at engineering students that focus on a systemic and control engineering curriculum. It encompasses fundamental topics along the complete process of modeling technical systems. Particularly, two major areas will be covered:

On the one hand, physical-based modeling techniques which derive formal model equations based on analyzing the physical first-principles of technical systems. This includes, inter alia, generalized equivalent circuits, bond graphs, port-Hamiltonian systems, variational analysis (Euler-Lagrange of the first kind). Selected topics of physical-based control methods will also be briefly introduced to integrate the complete physical control design in the wider control context and highlight its possible benefits.

On the other hand, data-based identification techniques will be covered which are used to identify concrete model parameters for a given technical system from experimental data sets. When combining the identification with an initial, non-physical, structural set up of model equations, the complete process is often referred to as data-based modeling or black-box modeling.

P. E. Wellstead: Introduction to Physical System Modelling
W. Borutzky: Bond Graph Methodology
A. van der Schaft, D. Jeltsema: Port-Hamiltonian Systems: An Introductory Overview
R. Isermann, M. Münchhof: Identification of dynamic systems : an introduction with applications

Course Material On Ilias all relevant course material (including lecture slides, exercise and tutorial sheets and semester schedule) can be downloaded

Attendance time in lecture/exercise: 60h

Preparation and revision of course content: 75h

Preparation and attendance of the oral exam: 30h

After finishing this lecture, the students

  • understand the general model concept as well as the characteristics of physical and data-based modeling and can describe their differences.
  • are able to structure complex systems and systematically analyze dependencies of subsystems.
  • are able to explain the general procedure of physical and data-based modeling, apply it to technical systems, and analyze the results.
  • are able to apply causal and non-causal modeling approaches and distinguish between them.
  • have gained an understanding of generalized, cross-domain, physical relationships and can develop models for electrical, mechanical, pneumatic and hydraulic systems. They can identify states and constraints.
  • can describe the relationship between generalized, cross-domain, physical models and basic procedures of physical-based control and explain their advantages / limitations based on basic knowledge of control engineering.
  • are able to explain different identification procedures for parametric models of static and dynamic systems, select, and apply appropriate procedures for given technical problems.
  • know basic procedures of learning-based identification and can describe their limitations.
  • students can estimate and judge the effects of disturbances and real conditions on the identification results.

Oral examination of approximately 20 minutes. The grades will be announced directly after the exam. In order to arrange exam appointments can be arranged at the secretary (Ms. Stassen). Beforehand, an online registration via CAS is mandatory!