Ruth Willet
Ruth, a third-year graduate student in Acoustics, is applying machine learning methods to analyze the Thames Water dataset, which contains over 38,000 acoustic recordings from more than 18,000 unique water pipeline sites across the UK. Her work aims to explore how machine learning can aid in acoustic prognostic health management (PHM) for early leak detection in water pipelines. Because most pipes are not leaking, the dataset is highly unbalanced, making it challenging to train accurate models. Ruth is testing and comparing various algorithms to see which ones perform best for fault detection, especially when compared to traditional signal processing techniques.