Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Universal Transient Stability Analysis: A Large Language Model-Enabled Dynamics Prediction Framework

Chao Shen, Ke Zuo, Mingyang Sun

Abstract

Existing dynamics prediction frameworks for transient stability analysis (TSA) fail to achieve multi-scenario "universality"--the inherent ability of a single, pre-trained architecture to generalize across diverse operating conditions, unseen faults, and heterogeneous systems. To address this, this paper proposes TSA-LLM, a large language model (LLM)-based universal framework that models multi-variate transient dynamics prediction as a univariate generative task with three key innovations: First, a novel data processing pipeline featuring channel independence decomposition to resolve dimensional heterogeneity, sample-wise normalization to eliminate separate stable or unstable pipelines, and temporal patching for efficient long-sequence modeling; Second, a parameter-efficient freeze-and-finetune strategy that augments the LLM's architecture with dedicated input embedding and output projection layers while freezing core transformer blocks to preserve generic feature extraction capabilities; Third, a two-stage fine-tuning scheme that combines teacher forcing, which feeds the model ground-truth data during initial training, with scheduled sampling, which gradually shifts to leveraging model-generated predictions, to mitigate cumulative errors in long-horizon iterative prediction. Comprehensive testing demonstrates the framework's universality, as TSA-LLM trained solely on the New England 39-bus system achieves zero-shot generalization to mixed stability conditions and unseen faults, and matches expert performance on the larger Iceland 189-bus system with only 5% fine-tuning data. This multi-scenario versatility validates a universal framework that eliminates scenario-specific retraining and achieves scalability via large-scale parameters and cross-scenario training data.

Universal Transient Stability Analysis: A Large Language Model-Enabled Dynamics Prediction Framework

Abstract

Existing dynamics prediction frameworks for transient stability analysis (TSA) fail to achieve multi-scenario "universality"--the inherent ability of a single, pre-trained architecture to generalize across diverse operating conditions, unseen faults, and heterogeneous systems. To address this, this paper proposes TSA-LLM, a large language model (LLM)-based universal framework that models multi-variate transient dynamics prediction as a univariate generative task with three key innovations: First, a novel data processing pipeline featuring channel independence decomposition to resolve dimensional heterogeneity, sample-wise normalization to eliminate separate stable or unstable pipelines, and temporal patching for efficient long-sequence modeling; Second, a parameter-efficient freeze-and-finetune strategy that augments the LLM's architecture with dedicated input embedding and output projection layers while freezing core transformer blocks to preserve generic feature extraction capabilities; Third, a two-stage fine-tuning scheme that combines teacher forcing, which feeds the model ground-truth data during initial training, with scheduled sampling, which gradually shifts to leveraging model-generated predictions, to mitigate cumulative errors in long-horizon iterative prediction. Comprehensive testing demonstrates the framework's universality, as TSA-LLM trained solely on the New England 39-bus system achieves zero-shot generalization to mixed stability conditions and unseen faults, and matches expert performance on the larger Iceland 189-bus system with only 5% fine-tuning data. This multi-scenario versatility validates a universal framework that eliminates scenario-specific retraining and achieves scalability via large-scale parameters and cross-scenario training data.
Paper Structure (44 sections, 32 equations, 7 figures, 12 tables, 2 algorithms)

This paper contains 44 sections, 32 equations, 7 figures, 12 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Preliminary and Problem Formulation
  3. Power System Model
  4. Transformer Attention Mechanism
  5. Methodology
  6. Data Processing
  7. Channel Independence
  8. Segmentation and Sample-wise Normalization
  9. Temporal Patch
  10. Augmented GPT Architecture for TSA
  11. Input Embedding
  12. Transformer Block Architecture
  13. Output Projection
  14. Two-Stage Fine-tune Strategy
  15. Freeze-and-Finetune Strategy
  16. ...and 29 more sections

Figures (7)

  • Figure 1: Comparison of attention mechanisms and iterative pblackiction in TSA-LLM.
  • Figure 2: Model structure and pipeline of TSA-LLM.
  • Figure 3: New England 39-bus system: The evaluation of rotor angle under the stable and unstable OC. The pblackicted $\delta$ by TSA-LLM ((c) and (d)) demonstrates close alignment with ground truth ((a) and (b)).
  • Figure 4: Representative $\delta$ trajectory comparison for SG 10 (39-bus system). Pblackictions and ground truths are plotted on the same axes for improved clarity.
  • Figure 5: The t-SNE visualization of stable/unstable sample feature maps for the proposed TSA-LLM and LSTM.
  • ...and 2 more figures