⚡ electrical engineering

Robust Current Regulation of MMC-based MTDC Power Systems based on Lyapunov Inequality

This paper proposes a robust, LMI-based static state-feedback controller for MMC-based MTDC systems that ensures fast transient response and stability under uncertainties while explicitly handling input saturation and overcurrent constraints, validated through CIGRE benchmark simulations in RTDS.

Original authors: Victor Daniel Reyes Dreke, Rahul Rane, Aleksandra Lekić

Published 2026-06-10

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Victor Daniel Reyes Dreke, Rahul Rane, Aleksandra Lekić

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the future of our power grid as a massive, high-speed delivery network. Instead of trucks on roads, we have electricity flowing through giant cables under the ocean and across continents. To move this energy efficiently, we use special "traffic controllers" called Modular Multilevel Converters (MMCs). Think of these MMCs as highly sophisticated traffic lights and speed governors for electricity.

The problem is that the road conditions are never perfect. Sometimes the wind changes, a cable gets damaged, or the demand for power suddenly spikes. These are the "uncertainties." If the traffic controller is too cautious, it moves too slowly, causing delays. If it's too aggressive, it might crash the system or overload the wires.

This paper introduces a new way to design these traffic controllers so they are both fast and safe, even when the road conditions are unpredictable.

The Core Challenge: The "Goldilocks" Problem

The authors explain that existing controllers usually have to choose between being fast or being safe.

The "Fast" approach: Like a race car driver who pushes the pedal to the metal. It gets you to the destination quickly, but if there's a sudden pothole (an uncertainty), you might lose control.
The "Safe" approach: Like a defensive driver who drives very slowly to ensure they never hit anything. You arrive safely, but it takes forever.

The goal of this paper is to build a controller that drives like a professional race car driver who also has a perfect safety system: fast enough to handle emergencies, but smart enough to never break the rules.

The Solution: A "Mathematical Safety Net"

The researchers used a mathematical tool called Lyapunov Inequalities (think of this as a sophisticated "safety net" or a "stability map").

The Map: They created a map of all the possible "bad weather" (uncertainties) the system might face, like changes in resistance or inductance in the wires.
The Safety Net: Using this map, they designed a controller that guarantees the system will stay within safe limits (like not letting the current get too high) no matter what the "weather" does.
The Optimization: Instead of just being safe, they used a special math trick (Linear Matrix Inequalities) to make the controller as aggressive as possible without ever touching the safety net. It's like finding the fastest route that stays exactly on the edge of the cliff without falling off.

How They Tested It

To prove their idea works, they didn't just do it on paper. They built a digital twin of a real-world power system (based on a standard model called the CIGRE benchmark) and ran it in a super-fast computer simulator called RTDS.

They pitted their new controller (let's call it the "RCR") against an existing, high-tech controller (the "OCR").

The Results:

The Crash Test: They simulated a major fault (like a lightning strike causing a short circuit).
The Outcome: Both controllers fixed the problem, but the new RCR was better at keeping the "traffic" (current) from getting too wild.
- When the system got stressed, the RCR kept the peak current lower (like keeping the car from swerving too hard).
- It recovered from the shock slightly faster.
- It wasted less energy during the recovery (less "overshoot").

The Bottom Line

The paper claims that by using this specific mathematical approach, they created a controller that is less conservative (less timid) than current methods. It doesn't just survive the chaos of the power grid; it handles it more smoothly and quickly, ensuring that the lights stay on and the wind farms keep sending power, even when things go wrong.

In short: They found a way to make the power grid's "traffic cop" faster and sharper, without ever letting it break the law.

Technical Summary: Robust Current Regulation of MMC-based MTDC Power Systems based on Lyapunov Inequality

Problem Statement
Modular Multilevel Converter (MMC)-based Multi-terminal High-Voltage DC (MTDC) systems are critical for future sustainable energy transmission, particularly for integrating offshore wind and large-scale solar resources. However, ensuring reliable operation requires precise current regulation under a wide range of uncertain conditions, including parameter variations, component tolerances, and external disturbances. The fundamental challenge lies in the control-theoretic trade-off: achieving fast transient response while maintaining robustness against uncertainties. Existing strategies often treat performance and robustness separately, using heuristic tuning for speed or synthesis methods like $H_\infty$ and $\mu$ -synthesis for robustness. While Linear Matrix Inequality (LMI)-based approaches offer robustness, they can be overly conservative, limiting transient performance. Furthermore, standard tuning of weighting matrices is time-consuming and does not systematically guarantee desired performance or constraint adherence.

Methodology
The paper proposes a design framework for a Robust Current Regulator (RCR) that synthesizes a static state-feedback gain using Lyapunov stability conditions and Linear Matrix Inequalities (LMIs). The methodology proceeds as follows:

System Modeling: The MMC dynamics are modeled using an average equivalent representation in the $dqz$ reference frame. The model accounts for structured uncertainties in passive components (arm and transformer inductance and resistance). The system is represented as an uncertain discrete-time state-space model. To facilitate zero-error tracking, the models for output current control (OCC) and circulating current control (CCC) are augmented to include state differences and input rates.
Control Architecture: The inner control loop utilizes a static state-feedback structure combined with a feedforward term. The feedforward gain is calculated based on nominal dynamics to provide steady-state voltage references, while the feedback gain ( $K$ ) regulates the error of the augmented state vector.
Synthesis via LMIs: The core contribution is the formulation of an optimization problem to determine the feedback gain $K$ $K$ . The objective is to maximize the domain of attraction (by minimizing the eigenvalues of the Lyapunov matrix $P$ $P$ , equivalent to maximizing $\det(P)$ $det (P)$ ) subject to:
- Lyapunov stability constraints for all admissible uncertainty realizations (ensuring closed-loop stability).
- State constraints (current limits) and input constraints (voltage reference limits).
- The constraints are defined as polytopic sets. The uncertainty set is treated as a convex hull of vertices.
Solution: Using Lemma 3.1 and Theorem 3.2, the non-convex constraints are transformed into a set of LMIs. The problem is solved using a semi-definite programming (SDP) solver (MOSEK). This approach explicitly accounts for system constraints without relying on tuned hyperparameters like $Q$ and $R$ matrices used in traditional optimal control.

Key Contributions

Less Conservative Synthesis: The paper introduces an LMI-based framework that synthesizes a static state-feedback gain which is less conservative than existing methods. It explicitly maximizes the constraint-admissible domain of attraction.
Constraint Integration: The design explicitly incorporates user-defined system constraints, including input saturation and overcurrent limits, ensuring that the regulator avoids excessively aggressive control actions while maintaining stability.
Robustness without Tuning: The method guarantees stability for all admissible uncertainty realizations without the need for time-consuming tuning of weighting filters or matrices.
Validation: The proposed RCR is validated against an Optimal Robust Current Regulator (OCR) from existing literature [11] using both MATLAB simulations with uncertain models and real-time simulations on the CIGRE MT-HVDC benchmark using RTDS®.

Results
The effectiveness of the RCR was assessed through two primary validation stages:

MATLAB Simulations: Simulations with 200 uncertain realizations of the MMC model showed that the RCR achieved a faster settling time (4.0 ms) compared to the OCR (4.3 ms). The RCR demonstrated a lower average Mean Absolute Error (MAE) relative to the nominal response ( $KPI_{RCR} \approx 446$ vs. $KPI_{OCR} \approx 714$ ), indicating behavior closer to the nominal case across uncertainties.
RTDS® Simulations: Using the CIGRE B4.57 DCS1 benchmark, the system was subjected to a symmetrical three-phase line-to-ground fault.
- Overcurrent Limitation: During fault clearance, the RCR limited the peak overcurrent to 2.25033 kA, whereas the OCR reached 2.51458 kA. Similarly, during post-fault recovery, the RCR limited the peak current to -6.51704 kA compared to -6.74151 kA for the OCR.
- Power Recovery: Both controllers restored power delivery to 200 MW with similar settling times (~0.23 s). However, the RCR resulted in a lower active power peak (5.628 MW) and significantly lower power excess (MAE of 77.84 MW vs. 79.65 MW for active power) compared to the OCR.
- Transient Performance: In the $dq$ reference frame, the RCR reduced peak overcurrents in both the output and circulating currents and achieved a 5% faster transient response in the $q$ -axis compared to the OCR.

Significance and Claims
The authors claim that the proposed method successfully resolves the trade-off between fast transient performance and robustness-induced conservativeness. By leveraging Lyapunov inequalities to synthesize the controller, the RCR provides the fastest transient response possible within the safe operation region for all possible uncertain realizations. The paper concludes that the proposed method offers a systematic way to design robust current regulators that outperform existing optimal robust methods in terms of tracking error and transient speed, while strictly adhering to physical system constraints. The authors note that this approach is generalizable to any linear state-space realization.

The Core Challenge: The "Goldilocks" Problem

The Solution: A "Mathematical Safety Net"

How They Tested It

The Bottom Line

More like this