Graph Signal Processing for Power Distribution System Monitoring
Over recent decades, reduction in the costs of advanced metering infrastructure (AMI) has improved the economic feasibility for these devices to be deployed throughout distribution systems. As a result, data on power systems has proliferated. This data has enabled the development of new data-driven algorithms for distribution system control. However,this recorded data is imperfect. Data recorded by AMI devices is prone to corruption.This data is also associated with potential unknowns in physical connectivity of the power distribution networks. These problems can impede the ability for a grid operator to successfully perform essential control tasks in power distribution systems. To this end, this thesis presents novel algorithms in the domains of bad data detection and network topology change detection. The developed algorithms are built upon the field of graph signal processing, which has received minimal attention in the context of power systems. This thesis additionally contributes towards the application of the graph signal processing algorithms in power distribution systems.