Research On Modeling Of Cyber-Physical Power System And Its Performance Analysis

With the development of digital communication,the operation of power grids is dependent on the assistance of communication networks increasingly.It means that the dual-network coupling has promoted the research on the cyber-physical power system(CPPS).As the basis of CPPS research,CPPS modeling is of great significance for extended research such as performance analysis.Therefore,exploring CPPS modeling and related performance analysis play a vital role in the construction and development of smart grids.Firstly,a more realistic model for CPPS topology is proposed.By analyzing the topological features of actual CPPS,its hierarchical structure and coupling relationship are constructed via the community overlapping and interdependent network.In case study,this model is applied to IEEE 39-bus and 118-bus systems,establishing the corresponding CPPS topologies.Their invulnerability is explored by the proposed coupled cascading failure.The results are compared with other topological models,which verifies the rationality and advantages of the proposed model.Then,a CPPS information flow model based on graph theory and traffic dynamics is proposed.This model takes into account the unique coupling of CPPS,the processing capability of communication equipment,and the transmission mode in CPPS.In case study,by applying the proposed model to the CPPS generated with different scales and structures,we explored the influences of network scale,critical nodes and topology structure on transmission efficiency,and the model effectiveness is attested.Finally,a CPPS vulnerability assessment model is proposed based on the Q-learning algorithm.In the context of topological attacks,Q-learning is used to train the attack strategy on CPPS topology model.The Q value is iterated from feedback rewards triggered by coupled cascading failure,building the learning framework for vulnerability assessment.The assessment model is applied to a CPPS model to investigate the convergence process and attack effect of Q-learning,and the correlation analysis is conducted.