Thesis subject
AI-powered anomaly detection in energy markets (MSc)
Have you ever wondered how energy markets operate under the energy transition? How can these markets be manipulated, and how can we effectively identify anomalies from energy market data? Are you interested in developing statistical, big data analytics, and AI solutions to identify such anomalies? Would you like to participate in a project to enhance the integrity of energy markets? If so, this thesis topic might be for you!
Short description
The Dutch energy system is undergoing a major transformation toward intermittent and unpredictable renewable energy sources like wind and solar. These intermittent energy sources could lead to economic and societal impacts, such as volatile energy prices when poorly integrated. Energy markets are essential for balancing supply and demand, but if not carefully managed, market anomalies may disrupt operations and price stability, either due to natural fluctuations or, at times, manipulative behaviour.
This thesis leverages statistical methods, big data analytics and/or AI to identify such anomalies within the Dutch energy market. Access to real-world datasets and tools from key stakeholders will support effective anomaly detection and data-driven insights, advancing energy market operations and supporting sound decision-making for the energy transition.
Students on this project will connect with leading researchers and may have opportunities for research visits/collaborations at top EU energy institutes, such as the Authority of Consumers and Markets (ACM) and the European Organization for Nuclear Research (CERN). For more information, you can check our HighLO project website.
Objectives
- Review articles and white papers on energy markets operation and anomaly detection methods.
- Exploring existing data-driven solutions/tools for analysing big data (e.g., energy markets trading data).
- Develop a data-driven model, supported by statistical solutions and/or AI, to detect anomalies in energy market data.
Tasks
The work in this master thesis entails:
- Read literature on energy market operations and anomaly detection methods, focusing on data-driven solutions for identifying market manipulations in energy markets.
- Develop and implement a data-driven AI model, using statistical methods as needed, to detect anomalies in energy market data and evaluate its effectiveness using real-world data from the Dutch energy market.
Literature
- Pang, Guansong, et al. "Deep learning for anomaly detection: A review." ACM computing surveys (CSUR) 54.2 (2021): 1-38. Link
- Scharff, Richard, and Mikael Amelin. "Trading behaviour on the continuous intraday market Elbas." Energy Policy 88 (2016): 544-557. Link
- Birkeland, D., AlSkaif, T., Duivenvoorden, S., Meeng, M., & Pennings, J. M.. " Quantifying and modeling price volatility in the Dutch intraday electricity market" Energy Reports 12 (2024): 3830-3842. Link
Requirements
- Courses: Programming in Python (INF-22306), Big Data (INF-33806), Statistics (MAT), Machine Learning (FTE-35306) or Deep Learning (GRS-34809)
- Required skills/knowledge: basic data analytics/statistics/machine learning and willingness to learn new data-driven/AI tools, interest in energy markets and sustainable energy transition
Key words: Data Analytics, Machine Learning, Statistics, AI, Sustainable Energy Transition, Energy Markets, Anomaly Detection
Contact person(s)
Tarek Alskaif (tarek.alskaif@wur.nl)
Kwabena Bennin (kwabena.bennin@wur.nl)