research20h
Model-Free Reinforcement Learning Control for Resilient Cyber-Physical Systems
This paper compares model-free controllers on a nonlinear system under cyberattacks, analyzing four RL reward types for accuracy, cost, and resilience. The Lyapunov reward offers the best resilience with low tracking error. Results inform builders on selecting RL controllers for resilient cyber-physical systems. The study provides insights into trade-offs between resilience, accuracy, and cost.
Key takeaways
- Lyapunov reward offers best resilience with low tracking error
- Exponential mode provides good trade-offs under moderate training
- Progressive and linear rewards converge faster but are less robust