Bachelor thesis / Master thesis: Development of a resource-saving machine learning-based condition monitoring system for planetary plain bearings based on component test data

The Chair for Wind Power Drives researches the behaviour of drive systems in modern multi-megawatt wind turbines. The research objectives are to increase the availability, robustness and energy efficiency of wind turbines and to reduce the levelised cost of energy. State-of-theart engineering software and system test benches are used for this purpose.

In order to reduce the power generation costs of modern wind turbines (WTs), the power density of the planetary gearboxes in WTs is to be increased. The use of plain bearings, which have already been used in industry for a number of years, is a suitable solution. To date, however, there is no realtime capable condition monitoring system (CMS) to predict and avoid critical operating points at an early stage. The CWD is therefore conducting research into a modern CMS network. Preliminary work has shown the suitability of individual sensors for detecting operating anomalies. Some sensors proved to be more suitable for certain operating conditions than others. In addition, the algorithms used to date are very resource-intensive and only have limited real-time capability and usability in real applications.

The idea for this student project is therefore the development of a resource-saving ML monitoring algorithm, to which measurement results from various sensors (SAW, AE, Acc, etc.) are transferred simultaneously and with the help of which operating anomalies are to be detected in real time. In the course of the work, targetoriented sensors are to be selected for the CMS.

Yor tasks:

• Research into the state of the art and the method
• Planning and execution of experiments on a component test bench
• Development of a resource-saving algorithm for condition monitoring of planetary plain bearings
• Selection of suitable sensors for different operating states

Requirements:

• Motivation to work independently and on one’s own responsibility, ability to communicate and work in a team, as well as a secure command of the German or English language
• Interest in wind energy, gear and plain bearing technology as well as machine learning
• Programming experience in Python desirable
• Previous knowledge in the field of machine learning desirable

What we offer:

• Scientific work in a highly motivated, interdisciplinary team
• Work on a topic with high industrial relevance
• Exciting combination of theory and practice
• Pleasant working atmosphere and intensive supervision
• Option to participate in a scientific publication
• Immediate start possible

We look forward to your application by email:

Tim Scholz, M. Sc. RWTH
Chair for Wind Power Drives

Campus-Boulevard 61, 52074 Aachen
tim.scholz@cwd.rwth-aachen.de