Multi intelligent agent based cyber attack resilient system protection and emergency control

Abstract

Centralized system protection and emergency control schemes are playing critical roles in maintaining the stability of modern power system. Being able to make optimal decisions based on global information and coordinate complex protection/control activities is the most desirable feature of a centralized scheme. On the other hand, centralized intelligence is quite vulnerable to various malicious cyber attacks, and the risk of losing it might be extremely high. This paper proposes a multi-agent based system protection and emergency control scheme with decentralized intelligence. The proposed decentralized scheme can detect coordinated cyber attacks with special patterns and carry out the best effort remedial actions in an adaptable manner, and such that system stability is maintained even part of the scheme is compromised. One decentralized load rejection scheme is designed and implemented in IEEE 39 bus system as a study case. Scheme performance is tested when under various attack vectors and desirable cyber attack resiliency is exhibited.