Olayiwola Arowolo

Olayiwola Arowolo

PhD Candidate, Delft AI Energy Lab @ TU Delft

Talks and presentations

Exploring Extrapolation of Machine Learning Models for Power System Time Domain Simulation

July 24, 2025

Conference talk, 12th Bulk Power System Dynamics and Control Symposium (IREP 2025), Sorrento, Italy

Abstract: Time domain simulation (TDS) is an important tool for assessing power system security under various disturbances. However, its computational cost limits the number of disturbances that can be assessed. The need for fast assessment of numerous disturbances has increased with the rapid integration of renewable energy sources. Machine learning (ML) methods have been explored to accelerate power system TDS, but these methods are studied in interpolation scenarios, where they predict outputs for inputs within the training data distribution. This work uses a state-of-the-art ML model to explore the extrapolation behaviour of ML models for TDS. First, we highlight the importance of ML models’ extrapolation capacity for fast assessment of numerous diverse disturbances. Next, we demonstrate that extrapolation for discrete disturbances is more challenging than for continuous disturbances. Subsequently, we investigate how transfer learning (TL) may be used to improve the performance of ML models in TDS extrapolation scenarios. Finally, we outline the limitations of TL for power system TDS and suggest alternative approaches for developing ML models with better extrapolation performance in TDS applications. Read the paper.

Predicting the Future Trajectory of Zebrafish with Probabilistic Machine Learning

September 23, 2022

Symposium talk, 1st Computation and Cognition Summer Internship Symposium, Tübingen, Germany

Abstract: Larval zebrafish is widely used in neuroscience research because it has a small transparent nervous system, allowing for cellular resolution examination of neural dynamics throughout its brain under a microscope. Recent advances in neural imaging have developed a microscopy technique that tracks the motion of freely swimming larval zebrafish and keeps its brain within the camera’s field of view. This technique utilizes model predictive control (MPC), which depends on the accurate prediction of the animal’s future position. However, larval zebrafish moves in discrete bouts making sudden, high-velocity movements that are difficult to predict. Without predictive control, tracking errors greater than 100 µm occur about 9% of the total time and 49% of the time when the fish is in motion. To reduce this error, we use a probabilistic gradient boosting prediction framework to determine the most likely region the fish can be found in the next 6 camera frames given a sequence of the fish’s coordinates and heading vector for the previous 10 frames. We use these predictions to plan the path of the tracking microscope. Compared to without prediction, we reduced the number of frames with tracking errors above 100 µm by an average of 40%. Our results demonstrate that, given a short history of previous positions, Machine Learning (ML) techniques can be used to predict the future trajectory of larval zebrafish by uncovering informative patterns in zebrafish locomotion. Watch the talk.