The objective of research project C is to enhance the monitoring and prediction of both prosumer behaviour and the state evolution of the underlying grid within the built environment. It will investigate methods of active computational learning to search large volumes of streaming sensor data for operator-defined events or automatic detection of statistically significant anomalies and develop new mathematical tools and algorithmic techniques for real-time monitoring and control of the smart grid infrastructure. The project is divided into two work packages.
The first package delivers new efficient procedures to search large volumes of streaming data for either specific or user-defined events or statistically significant anomalies. The development consists of an investigation of methods of active computational learning and results in a proposed solution which will help stakeholders to largely automate the process of data monitoring and event detection.
In the second package, new and advanced mathematical models and tools will be developed to handle large amounts of data to monitor, predict and control the (state) evolution of the grid infrastructure. One of the important contributions is the implementation of advanced methods based on the newly developed models on accelerators like GPUs and FPGAs.