Dynamic Flow Model Based Reliability Evaluation And Optimization Of Interdependent Networks In Smart Grid

With the rapid evolution of communication and information technology,the degree of intelligence and interconnection of contemporary power grid systems is continuously developing with unprecedented depth and breadth.The intelligent unified dispatching of the grid,flexible power transmission technology,and real-time two-way interaction with users enabled by this positive development will greatly optimize power flow distribution,reduce line losses and improve power load curves,thereby significantly improving the efficiency of power grid operations.At the same time,however,with the advancement of the intelligent and interconnected power grid system,the degree of interdependent coupling between the deeply integrated power transmission network and the power communication network is also continuing to deepen,bringing new challenges to the reliable operation of the power system.In recent years,large-scale power outages that have occurred at home and abroad have proved that this inter-network dependency,if no measures are taken to actively respond,will seriously aggravate the depth and breadth of the impact of grid cascading failures,and bring new potential risks to the stable operation of the grid.In view of the above practical needs and problems facing by the current power grid development,this thesis focuses on the reliability evaluation and optimization of the internal interdependent network of the smart grid system,and has achieved the following main research results:(1)This thesis takes the reliability assessment of the dependent network of the smart grid as the research starting point,and proposes a dynamic analysis model of the cascading failure of the interdependent networks of the power grid system based on dynamic power flow and communication data flow.The model proposed in this thesis leverages complex network theory to evaluate the structural reliability of the power-communication interdependent networks.At the same time,the actual operating characteristics of the power transmission flow and the communication data flow in the system are incorporated in the dynamic analysis of cascade failures.Thereby this proposed model is able to further reflect the impact of real-time changes in power transmission flow and communication data flow on the overall reliability of the power system in actual scenarios.The simulation results show that when the above dynamic characteristics are taken into account,the depth and breadth of the grid system cascading failures will show a deeper trend in general,and the trend and severity of failures propagation under different initial sudden failures will show greater differences.This difference further highlights the risks and criticalities of a small number of nodes and links in the power grid system during the development of cascading failures,and can provide an effective reference for further power grid system reliability optimization measurements.(2)Based on the proposed power-communication interdependent network dynamic cascading failure model,this thesis proposes a dynamic flow-based power-communication interdependent networks critical nodes identification and decoupling strategy to reduce the interdependence between the networks of power grid systems by averting potential further cross-layer failure propagations after the initial failure.Besides leveraging complex network theory to evaluate structural criticality of the internally coupling node pairs in the power-communication interdependent network,this strategy incorporates the dynamical operational mechanism of the power flow and the communication data flow into the node pair criticality assessment.It is verified by the simulation results of the dynamic cascading fault model proposed in this paper,that since the criticality analysis process of this proposed strategy incorporates the dynamic redistribution characteristics of the flows during the actual cascading failure process of the grid system,the proposed node decoupling protection strategy can further improve the reliability of the power-communication interdependent network in the face of cascading failures,compared to the benchmark strategies.(3)In this thesis,the power communication service routing problem is modeled as a Markov decision process model,and based on this model,an adaptive power communication service dual routing algorithm based on deep reinforcement learning is proposed,in order to realize the risk balance of power communication network through reasonable routing decisions,thereby suppressing the overall cascading failure risk of the power grid system.Compared with existing heuristic routing algorithms with pre-fixed strategy,the greatest advantage of the proposed algorithm is that it can be trained online to identify and execute the long-term optimal routing decisions and continuously re-adapt and re-optimize along with the time-varying status of power grid system.In addition,beyond considering the risks of the power communication network during the routing decision which is done by most previous researches,the proposed algorithm additionally considers the risks contained in the power transmission network that is highly coupled with the power communication network in power grid systems.This enables it to further realizes the overall risk of the power grid system.Finally,it is verified by the simulation results of the proposed power-communication interdependent networks dynamic cascading failure model,that the algorithm can further suppress the potential cascading failure risk due to power-communication interdependence compared with the benchmark algorithms,while maintaining reasonable network resource utilization.