面向智能电网的安全分布式频率控制关键技术研究

In recent years, attack events on power systems occur frequently and security has become the key factor in the evolution of the traditional power systems to smart grids. With the high penetration of distributed renewable energies, both of controllable generation power and the system inertia decrease. As a result, the traditional hierarchical frequency control framework is facing great challenges and the distributed frequency control strategy breaking the three-hierarchy is very promising. Once the distributed frequency control is intruded by the attacker, electronic devices damage and even blackout may happen to the smart grid. However, little attention has been paid to the security concern for the distributed frequency control. The project aims to develop the framework for the security of distributed frequency control. By taking into account the tight coupling of cyber space and physical space, the lack of the global information, local information exchanging and distributedly deployed physical devices, the project focuses on cyber-physical cooperative attack modeling, the active-passive and cross-layer combined attack detection algorithm and the resilient control method design with the guaranteed system performance.

近年来，各国电网安全事故频发，安全问题已经成为威胁智能电网建设的关键因素之一。随着新能源发电不断渗透，可控的发电比例和系统惯性减少，传统的基于调节发电的频率控制无法满足稳定性和可靠性需求。跨层的、协同负载和新能源发电的分布式频率控制具有可扩展性强、灵活性好等特点，使得其成为解决未来智能电网频率控制的核心技术。分布式频率控制遭遇攻击，将会导致智能电网中设备损毁、断电等事故。然而分布式频率控制的安全研究才刚刚开始。本项目以构建安全分布式频率控制的关键技术与理论为目标，充分考虑智能电网频率控制过程中信息网络与物理网络深度耦合、全局信息难以获取、时空数据分散但耦合等基本特征，重点解决信息物理协同攻击机理分析、融合跨层和主被动的分布式异常检测技术设计、保障系统性能的协同防御方法研究三个关键问题，同时对理论结果进行仿真及实验验证。

分布式控制由于可扩展性强、单点容错能力强等优势，是解决智能电网中分布式新能源发电渗透率增长问题的前沿关键领域。而由于分布式控制方法信息物理深度融合特性和智能电网复杂动态特性，保障智能电网包括频率控制在内的分布式控制应用安全性已成为信息领域最具有挑战性的课题之一。本项目综合考虑智能电网的控制目标、攻击入侵路径、信息物理融合特性，面向分布式控制性能指标，采用理论分析与实验验证相结合的方式，首先构建信息篡改、信息窃取和最优攻击模型，进而量化分析不同攻击对于系统性能的影响。在此基础上，结合智能电网的控制过程和电压电流动态特性，设计了基于移动目标防御的攻击检测和定位方法，研究了主动和被动攻击检测机制，同时基于网络拓扑冗余和分布式信息交互特点，研究了针对智能电网的典型分布式控制模型的弹性控制算法，针对智能电网的分布式控制建立了一套完整的“攻击建模与影响分析—攻击检测与定位—保性能的弹性控制”方法与理论体系。同时对理论研究成果进行了仿真和实验验证。