Enhanced resilience in smart grids: A neural network-based detection of data integrity attacks using improved war strategy optimization

Abstract

Ensuring the resilience and security of Smart Grid (SG) infrastructure is critical for sustainable energy management. This paper proposes a new probabilistic approach for identifying Data Integrity Attacks (DIAs), targeting decentralized consensus-based energy management algorithms. The method uniquely combines Artificial Neural Networks (ANNs) with an Improved War Strategy Optimization Algorithm (IWSOA) to determine optimal weight and bias factors, offering superior performance compared to existing techniques. Key advantages include: 1) it functions using only transmitted information and network topology, eliminating the need for private data access; 2) it is cost-effective and can be integrated into existing algorithm execution modules; 3) enhanced detection accuracy, achieving up to 99.5 % detection rate with 10 hidden neurons. The proposed framework demonstrates robust performance across various attack scenarios, effectively identifying DIAs in both single and multiple iterations. In a case study using the Future Renewable Electric Energy Delivery and Management (FREEDM) system, the method successfully detected 99.5 % of attacks that would have resulted in a 21 % profit increase for the attacker, thereby protecting the system's integrity. This approach significantly enhances SG infrastructure's resilience against DIAs, contributing to more secure and sustainable energy management.