A Distributed Optimal Dispatch Strategy for Virtual Power Plants Considering DoS Attacks Based on Neural Dynamic Optimization

다국어 초록 (Multilingual Abstract)

With the increasing penetration of distributed renewable energy (DRE) in power systems, the virtual power plant (VPP) has rapidly developed as a novel structure. However, due to the distributed nature and characteristics of the VPP, its spatially dispersed devices make it more vulnerable to cyberattacks. Against this background, a distributed neural dynamic optimization algorithm is fi rst proposed to solve the economic dispatch problem of the VPP. Then, considering the system instability under denial-of-service (DoS) attacks, the attack mechanisms and patterns of DoS on the VPP are analyzed.
For coordinated attacks launched by multiple attackers, the attack targets are refi ned, and diff erent defense strategies are adopted according to the degree of impact. For non-saturated DoS attacks, a self-fusing mechanism is designed to eliminate their infl uence. For nodes suff ering from saturated attacks, a reputation marginalization strategy is proposed to mitigate external threats. The stability of the control system under DoS attacks is proven via Lyapunov theory. The proposed method eff ectively resists DoS attacks while maximizing the retention of generating units within the VPP, enabling the VPP to operate at minimum cost under attack conditions. Finally, the eff ectiveness of the proposed strategy is verifi ed by a nine-node VPP system modeled in MATLAB/Simulink.