Proxel-Based Digital Twin for Reliability Assessment of Energy Systems through Data-Driven Fault Trees

  • Type:Master Thesis
  • Supervisor:

    Omar Mostafa

    Prof. Sanja Lazarova-Molnar

Description

Problem: 

Build a reliability-oriented digital twin for a small photovoltaic system. Automatically extract repairable/multi-state fault trees from operational logs (e.g., fault records, sensor data, weather) and simulate them using the Proxel-based method to obtain transient reliability/availability. Calculate component importance (e.g., Birnbaum or Fussell-Vesely) to prioritize maintenance and operational adjustments.  

Goal: 
Benchmark Proxel-based simulation results against Discrete-Event simulation in terms of accuracy and runtime, and evaluate sensitivity to data availability. The deliverables are a reproducible pipeline (Python), a case study report, and guidelines on data requirements for reliability-focused digital twins. 

Required Skills and Knowledge: 

Stochastic modeling (fault trees/Markov models); programming in Python; data visualization; power systems/PV fundamentals; simulation methods. Bonus: basic machine learning for simple classification.