Economic & Reliable DC Microgrids

直流微电网的网络物理系统(CPSS)环境由配备了网络能力的相互关联的主体组成，使能源和负载能够执行具有增强能力的危险任务。为了避免系统故障并提供可扩展的CPSS，迫切需要为CPSS开发有效的控制和可靠的调度策略。为了利用功率缓冲器将易变负荷和低惯性配电网解耦，将开发适用于直流微电网的分布式智能控制策略，使功率缓冲器能够在负荷突变时相互协助。将设计一种参与者-批评者结构，批评者逼近最优值函数，参与者学习分布式最优控制器。由此产生的控制律可以在线优化，并在每次负载变化时触发，使所提出的方法特别适合于安全关键设备。此外，对对手建模并分析他们的威胁级别使代理能够使用适当的缓解机制来预见潜在的威胁情况。在此基础上，使用自信和信任度量来检测连接的转换器的不当行为，以最小化受危害数据的传播，其中传入信息的可信性在每个具有基于信任的分布式控制器的转换器处计算。该项目包括严格的理论研究和广泛的验证，为多智能体CPSS设计具有弹性和高信度的控制协议，并将其应用于直流微电网。此外，拟议的经济可靠的直流微电网(ERDCM)项目管理计划将确保高效地交付拟议的项目，包括研究、培训、传播和开发。在这个项目结束时，这位研究员将成为一名国际研究领导者，研究概况、学术/国际联系以及人际交往能力都得到了极大的提高。

Cyber-physical systems (CPSs) environment for DC microgrids, composed of interrelated agents equipped with cyber capabilities, have enabled energy resources and loads to perform dangerous tasks with augmented capabilities. It is desired to develop efficient control and reliable scheduling strategies for CPSs to obviate system failures and provide scalable CPSs. To employ power buffers to decouple volatile loads and a low-inertia distribution network, distributed intelligent control policies for DC microgrids will be developed to enable power buffers to assist each other during abrupt load changes reciprocally. An actor-critic structure will be designed with critics approximating the optimal value function and actors learning the distributed optimal controller. The resulting control law can be optimized online and triggered at each load change, making the proposed approach particularly suitable for safety-critical devices. Furthermore, modeling adversaries and analyzing their threat levels empower agents with appropriate mitigation mechanisms to foresee potential threatening circumstances. On this basis, the misbehaviour of connected converters is detected using self-belief and trust metrics to minimize the propagation of compromised data, where the trustworthiness of incoming information is calculated at each converter with a trust-based distributed controller. This project includes rigorous theoretical studies and extensive verifications on designing resilient and high-confidence control protocols for multi-agent CPSs with applications to DC microgrids. In addition, the proposed Economic and Reliable DC Microgrid (ERDCM) project management scheme will ensure efficient delivery of the proposed project, including research, training, dissemination, and exploitation. At the end of this project, the Fellow will become an international research leader with research profiles, academic/international links, and interpersonal skills largely enhanced.