Artificial Intelligence-Based Secured Power Grid Protocol for Smart City

Area of Science:

• Electrical Engineering
• Computer Science
• Artificial Intelligence

Background:

• Modern power systems are increasingly complex, integrating numerous smart grid components that generate vast data volumes.
• Traditional computing struggles with smart grid data, necessitating AI-driven solutions for efficient management and security.
• Cyberattacks pose a significant threat to smart grid stability and efficiency.

Purpose of the Study:

• To develop an AI-based model using LSTM and RNN to enhance the dynamic properties of smart grids.
• To differentiate between normal system changes and real-time cyber threats, including those from Revised Encoding Schemes (RES).
• To propose a federated learning strategy for secure and efficient consumer power data sharing.

Main Methods:

• Development of LSTM and RNN models tailored to adaptively capture time-varying energy system attributes.
• Implementation of a federated learning strategy with edge cloud support for consumer data sharing, prioritizing privacy and communication efficiency.
• Design of optimization problems for Energy Data Owners (EDO) and energy service operations, considering non-independent and identically distributed (IID) data effects.

Main Results:

• The proposed LSTM RNN-based structure effectively identifies False Data Injection Attacks (FDIA) and other cyber threats.
• Simulations indicate the approach successfully encourages EDOs to use high-quality local models, boosting energy service provider payouts and reducing latency.
• LSTM models demonstrated longer training durations and higher training loss compared to other tested models, but achieved effective threat detection.

Conclusions:

• The developed AI-based method, particularly using LSTM RNN, provides a robust solution for detecting sophisticated cyber threats in smart grids.
• The federated learning strategy ensures consumer privacy while enabling efficient data utilization for improved grid operations.
• The approach enhances smart grid security, efficiency, and data management capabilities in the face of evolving threats and data complexities.