Data-Driven Predictive Control for Wide-Area Power Oscillation Damping

TL;DR

This work tackles damping of inter-area oscillations in grids interfaced by VSC-HVDC by evaluating data-driven predictive control (DPC) methods that bypass explicit system models. It compares DeePC, Transient Predictive Control (TPC), and Single-ARX controllers against industry-standard MR and LS damping approaches on a nonlinear four-area Kundur benchmark with an HVDC link, using offline data collection and online convex optimization. The results show ARX-based DPC (including TPC) delivers damping performance comparable to model-based methods while achieving significantly faster online computation (e.g., $\tau_f=60$ horizon with $m=3$ inputs and $p=3$ outputs, solved in under 1 ms with warm starts), and DeePC remains competitive though sometimes slower. A linearity analysis around the HVDC ports supports the use of LTI-based DPC despite nonlinear grid dynamics, indicating practical viability under realistic load variations and noise, though extreme scenarios (e.g., $75\%$ PSS deactivation) remain challenging. Overall, data-driven damping emerges as a promising approach for future grids with higher uncertainty and reduced modeling fidelity, enabling plug-and-play wide-area stabilization without full system models.

Abstract

We study damping of inter-area oscillations in transmission grids using voltage-source-converter-based high-voltage direct-current (VSC-HVDC) links. Conventional power oscillation damping controllers rely on system models that are difficult to obtain in practice. Data-driven Predictive Control (DPC) addresses this limitation by replacing explicit models with data. We apply AutoRegressive with eXogenous inputs (ARX)-based predictive control and its Transient Predictive Control (TPC) variant, and compare them with Data-enabled Predictive Control (DeePC) and two standard model-based controllers. The methods are evaluated in simulation on a system exhibiting both inter-area and local oscillation modes. ARX-based predictive control and DeePC both achieve effective damping, while the ARX-based methods require less online computation. Using warm-started, pre-factorized operator-splitting solvers, ARX/TPC control actions are computed in less than 1ms. These results demonstrate that DPC is a viable approach for power-system oscillation damping for the given test case.

Figures (14)

• Figure 1: One-line diagram of the four-area power system with HVDC link. The control inputs $(u^1, u^2, u^3)$ and outputs $(V, P, Q)$ are in red.
• Figure 2: Open-loop pole locations with varying amounts of active PSSs.
• Figure 3: Closed-loop with band-pass filtering and mean removal.
• Figure 4: Single-Excitation Doublet used to excite the system.
• Figure 5: Output responses corresponding to the input signals in \ref{['fig:linearity_base_input']}. The orange lines give the response when the $u^1$ Single-Excitation Doublet is applied ($u^2$ and $u^3$ are held constant at zero). The blue lines describe the $u^2$ Single-Excitation Doublet response and the black lines describe the $u^3$ Single-Excitation Doublet response.