Objective

Development of a machine learning model for automated fault detection in large PV system portfolios based on historical operating data.

Scope of Work:

1. Literature research on ML methods in PV fault diagnosis (e.g., LSTM, autoencoder, anomaly detection, …)
2. PV data storage management and analysis of operating data from approximately 500 PV systems
3. Development and training of an ML model for the detection of anomalies and performance losses
4. Validation and classification of the model using real PV shutdowns, fault cases, and failure events
5. Derivation of maintenance recommendations and automated action instructions

Your Profile

• Motivation and interest in the topic
• Background in machine learning
• Experience in python programming

Contact:

Franz Pernkopf (pernkopf@tugraz.at)