Dynamic State Estimation for Electrical Power Grids

Abstract

Increased electrification of society, as well as increased incidence of distributed power generation, causes more dynamic operation of modern electrical power grids. Methods for monitoring of power grids have traditionally focused on static operation but must in the future be expected to have to account for system dynamics. On the other hand, privacy concerns for individuals, commercial secrecy concerns for companies, and cost concerns for power system operators, may cause power systems to be only partly known and without sufficient sensors for traditional dynamic state estimation in power systems. This thesis contributes to the understanding of Simultaneous Input and State Estimation (SISE), in particular its stability properties, and develops methods for dynamic monitoring of partially known power systems based on SISE. The other main focus of the thesis is decentralized state estimation. Both the sheer scale of power networks, as well as the fact that different sections of the system may be operated by different companies, make a single centralized estimation center impractical and/or impossible. The thesis therefore develops decentralized estimation approaches for power systems, in which the observability requirements of existing decentralized estimation methods are relaxed. In addition, the thesis also addresses cybersecurity and fault detection as well as sensor placement for power systems and shows how SISE can be exploited in power system stabilizer (PSS) design.