Data-Driven Control For Power Oscillation Damping: Robustness and Closed-Loop Identification

We study data-based control algorithms for damping power oscillations through voltage source converter based high-voltage direct current transmission. Power systems are inherently time-varying due to the time dependent nature of dispatch as well as the connected load on multiple time scales. Due to the dynamic nature, conventional time invariant controllers need to be tuned conservatively to cope with the significant variations. A recently proposed direct data-driven predictive control algorithm, performs model-free optimal control. The overall aim of the paper is to bring this technology closer to a practical implementation. In particular, we improve robustness and demonstrate the good performance of the new closed-loop online-excited predictive control algorithm. This enhancement achieves good results with data collected in closed-loop operation without the insertion of additional excitation signals.