Review on Set-Theoretic Methods for Safety Verification and Control of Power System

Yichen Zhang, Yan Li, Kevin Tomsovic, Seddik M. Djouadi, Meng YuearXiv 2020 · arXiv

Abstract

As renewable energy sources introduce unprecedented uncertainty into power grids, traditional simulation-based safety assessments are reaching their limits. This paper reviews set-theoretic methods—mathematical frameworks that shift the focus from individual state trajectories to the analysis of entire sets of possible system states. By treating uncertainty as "unknown-but-bounded," these methods provide formal safety guarantees and define a "Region of Safety" (ROS) where the system is mathematically proven to remain secure. This review categorizes these methods into set operation-based and passivity-based approaches, discussing their trade-offs in computational complexity and precision.

Introduction: The Shift from Points to Sets

Traditional power system safety relies on Dynamic Security Assessment (DSA), which typically employs point-wise simulations like Monte Carlo methods to test specific "what-if" scenarios. However, simulation has a fundamental flaw: while it can prove a system is unsafe by finding a failure trajectory, it cannot definitively prove universal safety across the infinite possible paths generated by fluctuating wind and solar inputs.

Set-theoretic methods address this by calculating the "envelope" of all possible system trajectories. Instead of checking one path at a time, these methods ensure that the entire reachable set of states never intersects with an "unsafe set" (e.g., states leading to blackouts or equipment damage).

Methodological Frameworks

The authors categorize set-theoretic approaches into two primary mathematical families:

1. Set Operation-Based Methods

  • Lagrangian Methods: These track the evolution of specific geometric shapes—such as boxes, ellipsoids, or zonotopes—over time. They are computationally efficient for linear systems and are often used to "over-approximate" the reachable states of a system.
  • Eulerian Methods (Level Set Methods): These solve Hamilton-Jacobi partial differential equations (PDEs) to find the exact boundary of reachable states. While highly accurate for nonlinear systems, they suffer from the "curse of dimensionality," typically becoming impractical for systems with more than four continuous states.

2. Passivity-Based Methods (Barrier Certificates)

Instead of tracking the set itself, these methods search for a "barrier certificate"—a mathematical function (similar to a Lyapunov function) that acts as a boundary between safe and unsafe regions. If such a function exists, it proves that no trajectory starting in the "safe" set can ever cross into the "unsafe" set. These certificates are often found using Sum of Squares (SOS) programming.

Conceptual comparison of safety verification methods: (a) traditional point-wise trajectory simulation, (b) set-based reachability analysis (Lagrangian/Eulerian), and (c) barrier-based safety certificates (Passivity-based).

Key Insights: The Region of Safety (ROS)

A critical tool for grid operators is the Region of Safety (ROS). The ROS represents the largest set of initial conditions that guarantee safety regardless of the disturbance. If the current state of the grid is within the ROS, the operator has a mathematical guarantee of safety.

There is a fundamental trade-off between computational speed and "conservatism." Using simple shapes like boxes (Lagrangian) is fast but may suggest a system is unsafe when it actually isn't (over-approximation). Using complex polynomials (SOS) is more precise but harder to solve.

Comparison between the real Region of Safety ($X_I^*$) and the estimated ROS ($\bar{X}_I$) using barrier functions, illustrating the gap between mathematical proof and physical reality.

Results and Future Directions

Benchmark tests on a linearized single-machine infinite-bus (SMIB) system demonstrated that Lagrangian, Eulerian, and passivity-based methods produce highly consistent Regions of Safety. Lagrangian methods were noted for their efficiency in higher-order systems, while passivity-based methods provided semi-analytical expressions easier to integrate into real-time supervisory control logic.

Despite these successes, scalability remains the primary challenge. Real-world power grids involve thousands of nodes, far exceeding the current capacity of these formal methods. Future research focuses on distributed formal analysis—decomposing the grid into smaller, interconnected subsystems—and combining formal proofs with machine learning to handle complex, non-linear grid behaviors.

Summary updated

Abstract

Increasing penetration of renewable energy introduces significant uncertainty into power systems. Traditional simulation-based verification methods may not be applicable due to the unknown-but-bounded feature of the uncertainty sets. Emerging set-theoretic methods have been intensively investigated to tackle this challenge. The paper comprehensively reviews these methods categorized by underlying mathematical principles, that is, set operation-based methods and passivity-based methods. Set operation-based methods are more computationally efficient, while passivity-based methods provide semi-analytical expression of reachable sets, which can be readily employed for control. Other features between different methods are also discussed and illustrated by numerical examples. A benchmark example is presented and solved by different methods to verify consistency.
BibTeX
@misc{zhang2020review,
  author = {Yichen Zhang and Yan Li and Kevin Tomsovic and Seddik M. Djouadi and Meng Yue},
  title = {{Review on Set-Theoretic Methods for Safety Verification and Control of Power System}},
  year = {2020},
  eprint = {2001.00080},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2001.00080},
}
Figure