Progressive Clustering with Learned Seeds: An Event Categorization System for Power Grid.
Boyi Xie, Rebecca J. Passonneau, Haimonti Dutta, Jing-Yeu Miaw, Axinia Radeva, Ashish Tomar, Cynthia D Rudin · Software Engineering and Knowledge Engineering · 2012
Advances in computational intelligence provide improved solutions to many challenging software engineering problems. Software has long been deployed for infrastructure management of utilities, such as the electric power grid. System intelligence is in increasing demand for system control and resource allocation. We present a model for electrical event categorization in a power grid system: Progressive Clustering with Learned Seeds (PCLS) – a learning method that provides stable and promising categorization results from a very small labeled data. It benefits from supervision but maximally allows patterns be discovered by the data itself. We find it effectively captures the dynamics of a real world system over time.