Secure and scalable power system event identification with renewable integration via federated LSTM and adaptive privacy mechanisms
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
The power system event detection process must accelerate and become more precise with increasing penetration of renewable energy systems into the grid. The authors discuss a federated learning LSTM (FL-LSTM) that serves as a secure detection system for distributed grid operations, keeping user data private. The federated model achieves better cross-grid generalization through the integration of system physical constraints, which operate under a physics-guided loss function that unites swing-equation consistency with ROCOF limits and frequency nadir bounds. To address privacy and noise sensitivity, we implement an adaptive differential privacy mechanism that modulates Gaussian noise per event stream based on event frequency, maintaining a global ( ɛ , δ )-DP budget while preserving rare-event sensitivity. This facilitates improved event detection without compromising data privacy. Simulations on the modified IEEE 39-bus system with varying renewable levels show that, compared to benchmark LSTM using central SCADA/DC data, the federated model converges faster, identifies events more accurately, and requires less communication. It preserves its distributed nature, stays robust to unseen events, and proves to be a strong candidate for privacy-preserving event detection in renewable-rich power systems. • FL-LSTM enables secure event detection across distributed power grids. • Physics-guided regularization boosts cross-grid model compatibility. • Adaptive DP balances privacy and accuracy based on event frequency. • Outperforms centralized LSTM in accuracy and communication efficiency. • Robust to unknown events in renewable-rich power systems.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it