Mitigating Cyber Risks in Smart Cyber-Physical Power Systems Through Deep Learning and Hybrid Security Models

The rise of renewable energy integration in smart grids brings new cybersecurity challenges, prompting this study to examine vulnerabilities in Smart Cyber-Physical Power Systems (CPPS). The integration of renewable energy sources, such as wind and solar, into smart grids poses operational risks due to their decentralized and variable characteristics, particularly within the communication layers essential for real-time monitoring and control. While increasing integration of renewable energy sources does not directly impact cybersecurity vulnerabilities, the primary challenge arises from their decentralization. Addressing this decentralization requires the use of cyber layers between supply and demand, introducing vulnerabilities of cyber threats to the control and communication systems of the power system. These layers, vulnerable to diverse cyber-attacks like false data injection (FDI), denial of service (DoS), and replay assaults, might compromise grid stability and security. To address these risks, the research proposes a hybrid approach that integrates conventional cybersecurity strategies with machine learning (ML) approaches to improve cyber-attack detection. The research highlights the use of deep learning models, including Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks, for real-time anomaly identification in grid data. These models, developed using a PSCAD-simulated dataset augmented with synthetic cyber-attacks, exhibit considerable advancements in threat identification and mitigation. The study emphasizes the difficulties in identifying cyber risks in grids with significant renewable integration, such as frequency instability and diminished system inertia, and suggests energy storage alternatives and sophisticated forecasting models to mitigate these issues. By incorporating a novel pre-processing method that leverages feature derivatives, the proposed models achieve over 98% accuracy in detecting cyber threats, providing a robust framework for protecting smart power grids from evolving cyber risks.