Stud.IP Uni Oldenburg
University of Oldenburg
13.08.2022 23:07:55

Personal details

Title Early Failure Prediction of Wind Turbines with Machine Learning Tools

Condition based maintenance plays an essential role in further reducing the costs of wind energy. An effective maintenance isĀ  based on a reliable estimate of the state of wind turbines and an early prediction of possible failures. The project WiSA big data aims at developing corresponding data analysis tools based on a large and unique data set. This set consists of high frequency operating data of more than 100 wind turbines over three years plus corresponding alarmlogs and service data. A thesis could for example use neural networks in order to predict the temperature of a wind turbine component based on the operating data. Since the model will be trained with data of a healthy turbine, a large deviation of the model temperature from the measured one can then be used as an indicator for possible faults and failures. Other methods of interest could be anomaly detection with autoencoders or support vector machines.

Home institution Institute of Physics
Type of work practical / application-focused
Type of thesis Bachelor's or Master's degree
Author Dr. rer. nat. David Johannes Bastine
Status available
Problem statement

- Very good programming skills, preferably with Python

- Strong interest in applied mathematics and machine learning

- Ability to work independently

- Basic technical knowledge preferably of wind turbines

Created 21/07/22

Study data

Departments No departments assigned
Degree programmes
  • Bachelor's Programme Engineering Physics
  • Master's Programme Business Informatics
  • Master's Programme Physics, Engineering and Medicine
  • Bachelor's Programme Physics
  • Master's Programme Engineering Physics
  • Master's Programme Physics
  • Master's Programme Computing Science
  • Master's Programme Environmental Modelling
  • Bachelor's Programme Computing Science
  • Bachelor's Programme Physics, Engineering and Medicine
Assigned courses No courses assigned
Contact person