Multi-layer defense algorithm against deep reinforcement learning-based intruders in smart grids

The Internet of Energy envisions the next generation smart grids as a highly interconnected network, including advanced metering infrastructures, distributed energy resources, and bidirectional communication systems. open architecture IoE-based grid results in manifold security concerns, especially risk False Data Injection Attacks. attack may target technical aspects system since fabricating network's data misleads power scheduling routing strategies interrupts healthy operation system. Additionally, monetary motivation for intruder sometimes is main motivation. conventional cyber defense are unable to detect well-developed Attacks, particularly once takes advantage Deep Reinforcement Learning-based development framework that analyzes dynamic nature grids. This paper primarily outlines various possible passive attacks using statistical methods. Then, reinforcement learning-based an active generator developed, initialized by modeled attacks. algorithm can simulate network environment subsequently creates unclassified After creating attacker, multilayer developed Snapshot Ensemble Neural Network adoptable Auto Encoder known unknown threats. Performance evaluations real-world simulation prove proposed successfully both active, where accuracy false positive detection rate 98.82% 97.42%, respectively.