Research On Robustness Analysis And New Security Model Of Smart Grid Based On Cyber-Physical Convergence

The smart grid is the base that supports the rapid development of the entire country and is one of the most important critical infrastructures.The smart grid is Cyber-Physical Systems(CPSs)consisting of a power system and an information system.It has the characteristics of large scale,diverse node types,complex structure,complex spatio-temporal evolution,and multiple coupling relationships.These features work together to ensure the normal operation of the smart grid,as well as its function and robustness,such as cascading failures.The special position of the smart grid in the critical infrastructure network makes it being the target of malicious destructors,and it is also the primary object of the defender's protection.In all previous military confrontations around the world,critical infrastructure will be the target of the first round of attacks by both sides of the enemy,especially the power grids that provide electricity for the normal operation of various industries.In the age of peace,through attacking smart grid by criminals and terrorist organizations,it can also achieve more significant effects than other smart infrastructures,such as direct economic losses,factory shutdowns,public panic,traffic congestion,etc.While the informationization and intelligence of the power grids bring convenience,efficiency,energy saving and high profit to it,which also provide an attack platform and opportunity for attackers.The complexity of the attack strategy,the diversity of the attack methods,targeted attack,and the randomness of the attack time bring many difficulties and challenges to the defender.In 2015,the Ukrainian power grid blackout accident was a classic case of destroying the power grids through coordinated cyberattacks.In view of the robustness of the smart grid itself and the possible security threats,this thesis describes its network structure and characteristics,as well as various factors that affect its robustness and security.Its main research points are as follows:(1)In the research of critical node identification methods,the idea of Giant Efficiency sub-Graph(GEsG)is proposed and used to replace the Giant Component(GC)to characterize the performance of the entire network.The node efficiency loss is defined to verify the accuracy of critical node identification algorithm and to compare the advantages of GEsG and GC as evaluation criteria with node efficiency loss.The experimental results show that:(i)Whether top 10% or 20% of nodes are selected as critical nodes,critical degree is always better than other indicators and their accuracy are equal to 80%;(ii)GEsG is more in line with the actual efficiency situation of heterogeneous networks(power grids)than GC;(iii)The correlation coefficient value of critical degree(CD)and NEL is larger than the correlation coefficient value of degree,betweenness,closeness or degree of degree;(iv)a counter-intuitive conclusion: the more important nodes may not be generation nodes,but some load nodes.(2)In the research of the influence of different coupling relationships on the robustness of smart grids,a two-layer smart grid is constructed,and the power network and information network are divided into several autonomous subnets according to geographical location information.The Node Importance(NI)is defined to evaluate the impact of nodes on the network,and the Assortative Coupling in Subnets(ACIS)and Dissortative Coupling in Subnets(DCIS)and Random Coupling in Subnets(RCIS)are proposed based on NI of cyber and physical nodes.Through simulation experiments,the same conclusions were drawn from the two data sets of IEEE 118-Bus System and Italian high-voltage power transmission network:(i)DCIS applied to the top-down coupling link is more beneficial in enhancing the robustness of the smart grid against targeted attacks than other coupled modes.(ii)ACIS applied to the bottom-up coupling link is more beneficial in enhancing the robustness of smart grids against targeted attacks than other coupled modes.(iii)Network processing load capacity is positively related to network robustness.Increased tolerance can improve the robustness of interdependent networks.(3)From the research of cross-layer attack path exploration,(i)For the smart grid two-layer structure model,a forward backward exploration model is proposed,which comprehensively utilizes the threat propagation mechanism of the cyber layer and the failure diffusion mechanism of the physical layer to form real-time multiple reachable Cross-Layer Attack Paths(CLAPs).The threat propagation mechanism takes into account factors such as system vulnerabilities,the spread of viruses or commands,and time.The failure diffusion mechanism utilizes the power flow to trigger the load redistribution and then invalidate the targeted physical node.(ii)An evaluation method has been proposed to calculate the probability of cross-layer attack paths.The experimental results show that the cross-layer attack path of the smart grid can be generated accurately and timely.The defender can quickly find the most likely cross-layer attack path according to the generation probability.The cross-layer attack path is easily affected by the node state of the cyber layer and the node tolerance of the physical layer,which changes with time.(4)From the aspects of threat model and an integrated defense mechanism,(i)based on the knowledge background of attackers;three intrusion strategies in threat model are predicted such as Topology Structure Intrusion Strategy(TSIS)and Socio-economic Intrusion Strategy(SEIS)and Power Flow Intrusion Strategy(PFIS).TSIS focus on maximizing node loss;SEIS focus on maximizing the economic loss due to power node failures;but PFIS focus on the use of load-induced load redistribution to induce more node failures,leading to large-scale blackouts;(ii)For the above intrusion strategies,we proposed an integrated defense mechanism against three different intrusion strategies,to improve the robustness of the CPSs by classifying the cyber and physical nodes and changing the coupling relationship according to its protection level.The experimental results show that the integrated defense mechanism can not only hide the critical nodes and induce the attacker to select the unimportant nodes as the attack targets,but also improve the robustness of CPSs and enhance the system's ability to respond to coordinated cyber-physical attacks.In summary,this thesis studies the robustness and the new security model of the smart grid.From the background of targeted attack and CCPA,how to identify critical nodes,slow down the speed of cascading failure and improve robustness,and develop a new security model of an integrated defense mechanism against different intrusion strategies.From the aspects of promoting the robustness of smart grid,the new security model has strong theoretical significance and practical worthiness.