Supervisor: Helena Sax
Earliest start: immediately
Type: Master Thesis

The increasing integration of renewable energy sources challenges power system operation. Decentralized renewable energy generation lead to an increase in average transmission distances and a higher utilization of the transmission grid to balance geographical mismatches between supply and demand. Maintaining system security requires costly redispatch measures. Corrective redispatch offers a way to increase grid operation efficiency by allowing higher line loadings after contingencies, e.g. a line outage. To comply with pre-contingency line limits again, corrective actions, e.g. the adjustment of controllable assets after a contingency to relieve overloads, are taken. However, conventional centralized corrective redispatch decisions often suffer from communication delays.
Concept and Objectives
This work investigates local control laws for corrective (also known as curative) redispatch actions, which can provide faster system responses compared to centralized decision-making. Such control schemes can predefine local rules that determine corrective actions directly based on local measurements, ensuring rapid reaction without requiring communication. Building on previous research on corrective redispatch via power flow redirection using HVDC lines, this thesis will expand the work to include Grid Boosters, which are large-scale battery storage systems capable of relieving transmission lines during overloads. Integrating these assets introduces new degrees of freedom but also new dynamics, particularly regarding their state of charge and response time.
Research Questions

1. How can existing local control laws for HVDC redispatch be extended to include Grid Boosters?
2. What is the impact of fast-acting local control on transient stability and dynamic system behavior?

Research impact This work will provide insights into:

• Improved grid utilization without requiring immediate infrastructure expansion,
• Reduced redispatch costs, enabling more cost-efficient and sustainable grid operation,
• Enhanced stability and resilience through faster local corrective action.

Supervisor: Helena Sax
Earliest start: immediately
Type: Bachelor Thesis

Grid-following (GFL) converters are becoming an integral part of the modern power grid, as the share of Inverter-based Resources (IBRs) is increasing. Due to the low inertia of inverters, future grid control strategies must adapt to the changing landscape. Additionally, the integration of IT equipment opens up novel attack vectors for targeted manipulations. Hence, anomaly detection approaches become necessary to detect misbehaving inverters.
Goal: In a previous work, a MATLAB GFL converter model was developed at EINS. Given a converter parameter and grid configuration, the model simulates the converter behavior, measured at the PCC. Based on the measurements, a grid-code aligned classification approach then applies rules to detect anomalies. The goal of this thesis is to investigate the expressiveness of the anomaly features using Machine Learning (ML). Based on the simulated data, two models shall be trained, where one model uses the raw PCC measurements as input, while the other model uses the predicted anomaly features. Both models should predict the original converter parameters.
Based on the resulting model accuracy, quantify the effectiveness of the anomaly detection algorithm and its features, and propose possible improvements to the rule-set of the anomaly detection algorithm.