Convexity and Optimal Online Control of Grid-Interfacing Converters with Current Limits
New algorithm proves converter output regions are convex, enabling projected gradient descent with guaranteed convergence.
A team from the University of Washington (Lauren Streitmatter, Trager Joswig-Jones, and Baosen Zhang) has published a breakthrough control method for power electronic converters that interface renewable energy sources with the electrical grid. Their paper, "Convexity and Optimal Online Control of Grid-Interfacing Converters with Current Limits," proves mathematically that the feasible output region of any grid-connected converter is convex regardless of its filter topology. This fundamental insight allows them to formulate the control problem as a convex optimization, from which they derive a projected gradient descent controller with formal convergence guarantees—a significant advancement over traditional cascaded control loops that are difficult to tune and can destabilize during large grid disturbances.
The new controller operates in real-time, dynamically driving the converter toward optimal performance based on actual grid conditions rather than regulating around a fixed reference point. This is crucial as converter-based resources like solar inverters and battery systems proliferate, creating complex interactions. Simulation results demonstrate the controller's safe and stabilizing behavior in both simple single-converter-infinite-bus systems and more realistic multi-converter networks, all while strictly respecting the hardware's current limits. The work, available on arXiv under identifier 2603.17135, represents a shift from heuristic, structure-dependent control design to a principled, topology-agnostic optimization framework for power electronics.
- Proves the feasible output region of grid-connected converters is convex for any filter topology, a key theoretical foundation.
- Derives a projected gradient descent controller from a convex optimal control problem, providing formal convergence guarantees.
- Replaces complex, hard-to-tune cascaded loops with real-time optimization that adapts to grid conditions, maintaining stability during disturbances.
Why It Matters
Enables more stable, efficient integration of renewables like solar and batteries into the grid by fundamentally improving power electronic control.