Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A New Neural Network Paradigm for Scalable and Generalizable Stability Analysis of Power Systems

Tong Han, Yan Xu, Rui Zhang

TL;DR

The paper tackles the challenge of scalable and generalizable power-system stability analysis by introducing a neural stability descriptor built from aggregated NNs and a sample-augmented iterative training scheme. It demonstrates the framework on two stability paradigms: large-disturbance analysis via a neural energy function (NEF) and small-disturbance analysis via a neural decentralized stability condition (DSC), both designed for type-consistent power systems. The approach achieves increased generalizability across topology and parameter variations and reduces conservativeness compared to traditional analytical descriptors, while maintaining scalability. Numerically, the neural EF yields larger RoAs than analytical equivalents and the neural DSC provides less conservative stability regions than passivity-based DSC, with strong generalization across system sizes. The work suggests promising directions in meta-learning and broader descriptor forms to further enhance robustness and applicability in real-world grids.

Abstract

This paper presents a new neural network (NN) paradigm for scalable and generalizable stability analysis of power systems. The paradigm consists of two parts: the neural stability descriptor and the sample-augmented iterative training scheme. The first part, based on system decomposition, constructs the object (such as a stability function or condition) for stability analysis as a scalable aggregation of multiple NNs. These NNs remain fixed across varying power system structures and parameters, and are repeatedly shared within each system instance defined by these variations, thereby enabling the generalization of the neural stability descriptor across a class of power systems. The second part learns the neural stability descriptor by iteratively training the NNs with sample augmentation, guided by the tailored conservativeness-aware loss function. The training set is strategically constructed to promote the descriptor's generalizability, which is systematically evaluated by verification and validation during the training process. Specifically, the proposed NN paradigm is implemented for large-disturbance stability analysis of the bulk power grid and small-disturbance stability conditions of the microgrid system. Finally, numerical studies for the two implementations demonstrate the applicability and effectiveness of the proposed NN paradigm.

A New Neural Network Paradigm for Scalable and Generalizable Stability Analysis of Power Systems

TL;DR

The paper tackles the challenge of scalable and generalizable power-system stability analysis by introducing a neural stability descriptor built from aggregated NNs and a sample-augmented iterative training scheme. It demonstrates the framework on two stability paradigms: large-disturbance analysis via a neural energy function (NEF) and small-disturbance analysis via a neural decentralized stability condition (DSC), both designed for type-consistent power systems. The approach achieves increased generalizability across topology and parameter variations and reduces conservativeness compared to traditional analytical descriptors, while maintaining scalability. Numerically, the neural EF yields larger RoAs than analytical equivalents and the neural DSC provides less conservative stability regions than passivity-based DSC, with strong generalization across system sizes. The work suggests promising directions in meta-learning and broader descriptor forms to further enhance robustness and applicability in real-world grids.

Abstract

This paper presents a new neural network (NN) paradigm for scalable and generalizable stability analysis of power systems. The paradigm consists of two parts: the neural stability descriptor and the sample-augmented iterative training scheme. The first part, based on system decomposition, constructs the object (such as a stability function or condition) for stability analysis as a scalable aggregation of multiple NNs. These NNs remain fixed across varying power system structures and parameters, and are repeatedly shared within each system instance defined by these variations, thereby enabling the generalization of the neural stability descriptor across a class of power systems. The second part learns the neural stability descriptor by iteratively training the NNs with sample augmentation, guided by the tailored conservativeness-aware loss function. The training set is strategically constructed to promote the descriptor's generalizability, which is systematically evaluated by verification and validation during the training process. Specifically, the proposed NN paradigm is implemented for large-disturbance stability analysis of the bulk power grid and small-disturbance stability conditions of the microgrid system. Finally, numerical studies for the two implementations demonstrate the applicability and effectiveness of the proposed NN paradigm.

Paper Structure

This paper contains 41 sections, 1 theorem, 28 equations, 11 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Proposed Neural Network Paradigm
  3. Framework
  4. Construction of Neural Stability Descriptors
  5. Power system decomposition
  6. Neural network design
  7. Rule-invariant operations
  8. Sample-Augmented Iterative Training Scheme
  9. Training set
  10. Model Training
  11. Verification
  12. Validation
  13. Sample augmentation
  14. Large-Disturbance Stability Analysis
  15. Modeling of System Dynamics and Energy Function
  16. ...and 26 more sections

Key Result

Theorem 1

The system (eq-8-1-3) is asymptotically stable iff all generalized eigenvalues of the pair $(\bm{E}, \bm{A}(\bm{\rho}) )$ exhibit a negative real part, i.e, $\lambda_{\max}(\bm{E}, \bm{A}(\bm{\rho})) < 0$ with $\lambda_{\max}(\cdot)$ being the largest real part of the generalized eigenvalues of the

Figures (11)

  • Figure 1: Framework of the proposed NN paradigm.
  • Figure 2: Schematic diagram of the construction of neural stability descriptors.
  • Figure 3: Illustration of two different construction strategies for the training set.
  • Figure 4: Illustration of the neural EF.
  • Figure 5: Illustration of the training set generation.
  • ...and 6 more figures

Theorems & Definitions (3)

  • Definition 1: Type-consistent power systems, system instance
  • Definition 2: Energy function
  • Theorem 1