IDP at the Institute for Machine Tools and Industrial Management
The iwb offers exciting interdisciplinary projects (IDPs) in deep learning, digital twins, predictive maintenance, and cyber-physical systems for computer science students. Together with a research assistant, a project topic is worked out, which also includes application-oriented and hardware-related work on the machines in the iwb's laboratory.
In addition to the project, students will attend an accompanying course at the iwb of their choice.
The iwb also offers student assistants positions and thesis topics for computer science students.
The following list includes the offered topics with their areas of application. If you are interested, please do not hesitate to contact the corresponding research assistant:
Web-Based Cost Model for a Lithium-Ion Battery Gigafactory
Starting Point
The cost-efficient production of high-performance lithium-ion batteries is one of the biggest challenges in the market penetration of electromobility. To increase competitiveness, it is crucial to understand the cost structure of lithium-ion battery production. This requires models that provide a comprehensive overview of a gigafactory.
Objective
Within this project, an existing web model for the full cost calculation of lithium-ion batteries will be optimized and extended with additional functionalities. The structure of the backend and frontend is already defined, so the project will focus on the step-by-step extension of the model and the website. Students are welcome to get involved in the architecture of the project and contribute their own ideas. A detailed documentation should be provided at the end to ensure that the website can be maintained and extended sustainably.
Requirements Profile
- Interest in battery cell production
- Understanding of technical processes
- Knowledge of Python / MongoDB
- Knowledge in web development
Recommended Courses
- Lithium-Ion Battery Production
- Industry 4.0
Contact
Maximilian Lechner
Maximilian.Lechner(at)iwb.tum.de
Use of Deep Learning in Joining Technology
Friction stir welding (FSW) is a solid-state welding process, which is applied in the aerospace sector. FSW is also considered an enabler for lightweight construction in electromobility. However, the demand for high-quality products and cost-effective production increases the need for inline process monitoring, which is being investigated at iwb. A meaningful and valid evaluation of the process variables could supplement or substitute other tests that follow welding. Time series data (process forces, temperatures, ...) and also camera data can be used for inline process monitoring. Various methods of data processing, from statistics to machine learning approaches, are used in a practical way to detect patterns and correlations between the recorded data and process variables and effects. Students of informatics have the opportunity to apply their knowledge in the field of information technology and to develop their expertise in an important field of production technology.
M.Sc. Fabian Vieltdorf
Data analysis and machine learning for monitoring drilling processes
In aircraft production, structural components are often joined using rivets. This requires drilling holes through multi-layer material packages. This can lead to manufacturing defects, such as increased burr formation. The goal is to detect these defects in real time based on sensor data.
As part of an IDP project, recorded sensor data can be analyzed using statistical methods or machine learning to investigate its correlation to manufacturing defects. Alternatively, extending the data pipeline, improving the database, implementing a simple process simulation, or implementing a user interface are also possible.
charlotte.winkler(at)iwb.tum.de
Implementation of Digital Twins for Machine Tools, Industrial Robots and Machining Processes
A digital twin is a machine tool’s virtual representation that accompanies the system throughout its lifecycle. It can represent the current behavior of the machine and can be used to predict its future state. Research foci at the iwb lie in the software-based modelling and visualization of the virtual launch, the data-based adaptation of the virtual machine model due to time-varying aspects of the physical machine and cloud-based applications to monitor its state. Within the framework of various IDPs, algorithms and methods that increase the manageability, accuracy and predictive power of the digital twin are analyzed, developed and applied.
M.Sc. Jannik Hüllemann
jannik.huellemann(at)iwb.tum.de
Predictive Maintenance in Production Planning and Control
The use of predictive maintenance holds great potential for avoiding machine failures and optimizing the production process. At iwb, artificial intelligence is used to predict the remaining useful life of components and this information is used to schedule production and maintenance in an integrated manner. Statistical methods, machine learning methods as well as classical mathematical optimization approaches are used to predict the remaining useful life. Relevant for the use of these algorithms are also the construction of data pipelines for reading and storing the sensor readings as well as the implementation of user interfaces for configuring the algorithms and displaying the results. An IDP can also cover several topics if required and of interest.
Cyber-Physical Systems in Assembly Technology and Robotics
In industry and at the iwb, topics in the area of digitalization, communication technology, and decentralized production systems have played a decisive role for several years. These topics are especially necessary for reconfigurable production systems, as companies have to develop innovative solutions to adapt to a high number of variants and increasingly customized products. This is also the context for the Plug&Produce concept, which takes approaches in the field of automated planning, commissioning, and monitoring of production systems into account. In this course various IDPs (Interdisciplinary Internships) dealing with concepts and implementations in the following subject areas are available: intelligent data processing, decentralized communication architectures (e.g., OPC UA), automated assembly/plant planning, and task/skill-oriented programming.
M.Sc. Stephan Trattnig