Machine Learning in Engineering - Theoretical FoundationsModule description #51147 / Version 1
- 6 CP
- Oral exam
- Graded
- Summer semester
- German
| Module / Version | #51147 / #1 |
|---|---|
| Validity | Since Summer semester 2025 |
| Available language(s) | English, German |
| Credit points | 6 |
| Person responsible for the module | Klinge, Sandra |
| Grading | Graded |
| Exam format | Oral exam |
| Language(s) of instruction | German |
Affiliation
| Faculty | Fakultät V |
|---|---|
| Institute | Institut für Mechanik |
| Academic chair | 35371300 FG Mechanik, insbes. Strukturmechanik |
Contact
| Office | Not specified |
|---|---|
| Contact person | Not specified |
| E-mail address | Not specified |
| Website | Not specified |
Learning outcomes
In engineering, machine learning methods are being used increasingly and in growing variety to accelerate and stabilise time-consuming numerical calculations, to make extensive measurement data usable, as material models and in many other combinations to accelerate or simplify working methods.
This module teaches the theoretical and mathematical foundations of machine learning and selected applications in mechanics. On the one hand, this includes automatic differentiation with dual numbers, optimisation algorithms for neural networks and other machine learning architectures currently being researched, regularisation methods, quality considerations for multivariate regression learning and possible combinations for machine learning methods. On the other hand, non-linear kinematics, the mechanical equilibrium condition as a PDE as well as non-linear material behaviour and the concept of polyconvexity are dealt with. The focus is on an in-depth understanding of mathematical and physical concepts and preparation for responsible method selection, application and implementation.
After successfully completing the module, students have the following knowledge
Knowledge:
- Knowledge of essential machine learning methods and techniques including relevant algorithms
- Understanding of the structure and functionality of neural networks and other machine learning architectures
- Knowledge of non-linear kinematics and the mechanical equilibrium condition
- Knowledge of non-linear material models
Skills:
- Selection, implementation and optimisation of neural network architectures
- Selection and implementation of regularisation methods
- Calculation and implementation of non-linear kinematics
- Implementation of conventional non-linear material models
- Derivation and treatment of the mechanical equilibrium condition
- Appropriate visualisation of the results
Competences:
- Independent use and development of suitable machine learning techniques
- Modelling of non-linear material behaviour using machine learning methods
- Solving partial differential equations using physics-informed machine learning techniques
This module teaches the theoretical and mathematical foundations of machine learning and selected applications in mechanics. On the one hand, this includes automatic differentiation with dual numbers, optimisation algorithms for neural networks and other machine learning architectures currently being researched, regularisation methods, quality considerations for multivariate regression learning and possible combinations for machine learning methods. On the other hand, non-linear kinematics, the mechanical equilibrium condition as a PDE as well as non-linear material behaviour and the concept of polyconvexity are dealt with. The focus is on an in-depth understanding of mathematical and physical concepts and preparation for responsible method selection, application and implementation.
After successfully completing the module, students have the following knowledge
Knowledge:
- Knowledge of essential machine learning methods and techniques including relevant algorithms
- Understanding of the structure and functionality of neural networks and other machine learning architectures
- Knowledge of non-linear kinematics and the mechanical equilibrium condition
- Knowledge of non-linear material models
Skills:
- Selection, implementation and optimisation of neural network architectures
- Selection and implementation of regularisation methods
- Calculation and implementation of non-linear kinematics
- Implementation of conventional non-linear material models
- Derivation and treatment of the mechanical equilibrium condition
- Appropriate visualisation of the results
Competences:
- Independent use and development of suitable machine learning techniques
- Modelling of non-linear material behaviour using machine learning methods
- Solving partial differential equations using physics-informed machine learning techniques
Content
- Historical development of AI, categorisation and sub-areas of machine learning
- Overview of application areas of ML in mechanics
- Neural networks, autograd, optimisers, hyperparameters, regularisation techniques
- Other machine learning algorithms (e.g. Kolmogorov-Arnold network, transformer)
- Nonlinear continuum mechanics
o Large strains
o Mechanical equilibrium condition as PDE
o Hyperelasticity, polyconvexity, plasticity
- Overview of application areas of ML in mechanics
- Neural networks, autograd, optimisers, hyperparameters, regularisation techniques
- Other machine learning algorithms (e.g. Kolmogorov-Arnold network, transformer)
- Nonlinear continuum mechanics
o Large strains
o Mechanical equilibrium condition as PDE
o Hyperelasticity, polyconvexity, plasticity
Module components
Compulsory area
The following courses are mandatory for this module:
Workload and credit points
Machine Learning in Engineering - Theoretical Foundations (Lecture)
| Workload description | Multiplier | Hours | Total |
|---|---|---|---|
| Attendance | 15 | 2 h | 30 h |
| Pre/post processing | 15 | 4 h | 60 h |
| 90 h | |||
Machine Learning in Engineering - Theoretical Foundations (Project)
| Workload description | Multiplier | Hours | Total |
|---|---|---|---|
| Attendance | 15 | 2 h | 30 h |
| Pre/post processing | 15 | 4 h | 60 h |
| 90 h | |||
The workload of the module adds up to 180 hours. The module therefore comprises 6 credit points.
Description of teaching and learning methods
- Lecture with blackboard and projections; explanation of theoretical principles and solution methods
- Project meeting
- Consultation with the supervisor
- Independent work, self-study
- Submission of a final report
- Project meeting
- Consultation with the supervisor
- Independent work, self-study
- Submission of a final report
Requirements for participation and examination
Desirable prerequisites for participation in the courses:
Numerical Implementation of linear FEM
Introduction to the FEM
Fundamentals of numerics (e.g. differential equations and numerics for mechanical engineering)
Fundamentals of analysis and linear algebra (e.g. Analysis I for Engineers, Analysis II for Engineers, Linear Algebra for Engineers)
Programming skills (e.g. in Octave/ Matlab/ Python/ ...)
Introduction to the FEM
Fundamentals of numerics (e.g. differential equations and numerics for mechanical engineering)
Fundamentals of analysis and linear algebra (e.g. Analysis I for Engineers, Analysis II for Engineers, Linear Algebra for Engineers)
Programming skills (e.g. in Octave/ Matlab/ Python/ ...)
Mandatory requirements for the module examination registration:
This module has no examination requirements.
Completion of the module
| Grading | Graded |
|---|---|
| Exam format | Oral exam |
| Language | English, German |
| Duration/Extent | 20 minutes |
Duration of the module
The following number of semesters is estimated for taking and completing the module: 1 semester(s).
This module can be started in the following semesters: Summer semester.
Maximum number of participants
The maximum capacity of participants is 60.
Registration procedures
none
Recommended reading, lecture notes
Lecture notes
unavailable
Electronic lecture notes
available
Recommended literature
- J. Bonet, R. D. Wood: Nonlinear Continuum Mechanics for Finite Element Analysis. Second edition, Cabridge University Press, 2008.
- O. C. Zienkiewicz, R. L. Taylor, J. Z. Zhu: The Finite Element Method: Its Basis and Fundamentals. Butterworth-Heinemann, 2013.
- S. Kollmannsberger et al.: Deep Learning in Computational Mechanics - An introductory course. Springer, 2021.
Assigned programs
This module version is used in the following programs:
| Program / program version | Semester(s) | Curriculum component |
|---|---|---|
| Computational Engineering Science (Informationstechnik im Maschinenwesen) StuPO 2018 (17.01.2018) | SS 2025 | 1. Kernbereich 1: Informatik und Mathematik |
| WS 2025/26 | 1. Kernbereich 1: Informatik und Mathematik | |
| SS 2026 | 1. Kernbereich 1: Informatik und Mathematik | |
| WS 2026/27 | 1. Kernbereich 1: Informatik und Mathematik | |
| Engineering Science StuPO 2007 (19.12.2007) | SS 2025 | 1. Mathematische Methoden |
| WS 2025/26 | 1. Mathematische Methoden | |
| Engineering Science StuPO 2020 | SS 2025 | 1. Mathematische Methoden |
| WS 2025/26 | 1. Mathematische Methoden | |
| SS 2026 | 1. Mathematische Methoden | |
| WS 2026/27 | 1. Mathematische Methoden | |
| Mechanical Engineering StuPO 2017 | SS 2025 | Wahlpflichtmodule › Profilmodule › Informationstechnische und rechnergestützte Modellierung |
| WS 2025/26 | Wahlpflichtmodule › Profilmodule › Informationstechnische und rechnergestützte Modellierung | |
| SS 2026 | Wahlpflichtmodule › Profilmodule › Informationstechnische und rechnergestützte Modellierung | |
| WS 2026/27 | Wahlpflichtmodule › Profilmodule › Informationstechnische und rechnergestützte Modellierung | |
| Vehicle Engineering StuPO 2017 | SS 2025 | 2. Profilmodule › 2.3 Ingenieurtechnische Grundlagen und Methoden |
| WS 2025/26 | 2. Profilmodule › 2.3 Ingenieurtechnische Grundlagen und Methoden | |
| SS 2026 | 2. Profilmodule › 2.3 Ingenieurtechnische Grundlagen und Methoden | |
| WS 2026/27 | 2. Profilmodule › 2.3 Ingenieurtechnische Grundlagen und Methoden |
Miscellaneous
Not specified