Predictive Maintenance for Critical Infrastructure

Potential supervisors: 
Research groups/keywords: 

*** Motivation ***

Cyber-physical automation processes operating in critical infrastructures typically involve expensive and safety-critical equipment and machinery. Oftentimes, these systems have real-time availability and uptime requirements, which makes unexpected failures of sensitive machines highly costly and in some cases dangerous to society. Predictive maintenance is one effective way of remedying this situation by continuously monitoring and analyzing available machinery data to detect early signs of degradation or failures. In this master thesis, the students will work closely with Vattenfall to explore the possibility of real-world application of predictive maintenance techniques.

*** Challenge ***

In industrial environments, the data to be analyzed is often not of a good quality for various reasons. It is therefore a challenge to preprocess data so that it is clean, structured, representative, and relevant to the desired task. Once a clean dataset is produced, the task is to apply and compare state-of-the-art predictive-maintenance algorithms. The expected outcome of this thesis is to draw conclusions pertaining to the feasibility, practicality, and efficacy of predicting imminent machine failures based on the performed analysis.

*** Background & Requirements ***

  • Computer Science or related programs
  • Machine learning and/or data mining
  • Embedded systems
  • Knowledge about network protocols

*** How to Apply? ***

Please send the supervisor Wissam Aoudi an email during fall 2019. The thesis will be set when we find suitable candidates.

Date range: 
October, 2019 to October, 2024