Power System Identification and Online Tuning of Power System Stabilizers

In this master thesis a framework for the identification of a low-order model of the electric power system and the online tuning of Power System Stabilizers (PSSs) is presented. The goal is to improve the damping of power oscillations, i.e., active power oscillations that emerge due to the interaction between the synchronous generators present in a power system. If these oscillations are not dampened through proper control action, they can become problematic, increasing the possibility of a total power system collapse. The most used controller to damp these oscillations is the PSS, which is installed in some of the large electric generators present in the grid. The standard approach in the field is to deploy these controllers following a "set and forget" approach, meaning that the controllers are tuned once during the initial power plant commissioning, and almost never re-tuned during the lifetime of the generator. However, in modern power systems the configuration of the system is not static, but continuously changes over time, for example due to the installation of new power sources or due to changes in grid topology. Even during normal operation there may be large variability, for example due to different power generation profiles yielded by renewable sources. Hence, in order to address these issues, the PSSs should be ideally adapted to the new configuration in order to guarantee proper oscillation damping performance. To solve this problem, we propose a solution based on a system identification procedure to first identify a low-order linear system of the power system. Then we use this estimated model to tune the PSSs on the current configuration. To validate the proposed method, we perform simulations in Matlab and in Hypersim, a real-time power system simulator.