Publikation
Decentralized Data-Driven Tuning of Droop Frequency Controllers
Allan Almeida Santos; Edwin Mora; Jan Peters; Florian Steinke
In: IEEE PES Innovative Smart Grid Technologies Europe. IEEE PES Innovative Smart Grid Technologies Europe (ISGT-Europe-2020), October 26-28, Delft, Netherlands, Pages 141-145, IEEE, 2020.
Zusammenfassung
Power oscillation damping is an important task for maintaining electric grid stability and reliability. It is realized via several control mechanisms that are traditionally tuned manually at installation time, often with the help of dynamic simulations of the full grid. With the increasing number of Decentralized Energy Resources (DER) and variable renewable energies, both the grid structure as well as generators' parameters are subject to frequent changes, turning manually tuned controllers suboptimal and existing dynamic models invalid. We therefore develop an automated controller tuning approach for the core mechanism of droop frequency control. It runs decentrally on each generator or in each control area without using an a priori dynamic model of the grid. Instead, we use data-based system identification to estimate a dynamic model of the remaining grid and combine it with local parameters to optimally adapt the local droop value. In simulation experiments, the algorithm is compared against an optimization approach, which has complete knowledge of the full grid configuration, and a heuristic that simulates manual tuning. In the considered setup, our approach reduces frequency oscillations by 15% compared to simulated manual tuning and comes within 8% of the all-knowing central approach.