Abstract
This brief focuses on improving the load frequency control (LFC) performance of a multi-area power system compromised by covert attacks. In this regard, a novel detection technique based on the multiplicative watermarking and Luenberger observer is developed to identify covert attacks applied to the system. In the presented detection strategy, a special watermark generator filter produces a dummy signal on the original signal transmitted through the insecure communication channels. In addition, there is a watermark equalization filter at the end of the communication channel, which cancels out the impact of the watermark signal, resulting in the recovery of the original signal. A PI controller tuned by the deep reinforcement learning algorithm and dependent on the Luenberger observer is utilized as the secondary frequency control strategy to mitigate the impacts of the convert attacks. The OPAL-RT real-time emulator-based results show the proposed defense strategy successfully detects and mitigates the covert attacks imposed on the system.
| Original language | English |
|---|---|
| Journal | IEEE Transactions on Circuits and Systems II: Express Briefs |
| Volume | 71 |
| Issue | 10 |
| Pages (from-to) | 4576-4580 |
| Number of pages | 5 |
| ISSN | 1549-7747 |
| DOIs | |
| Publication status | Published - 2024 |
Keywords
- Computer crime
- Frequency control
- Generators
- Load frequency control
- Luenberger observer
- Observers
- Power system stability
- Vectors
- Watermarking
- covert attacks
- deep reinforcement learning
- multi-area power system
- multiplicative watermarking