Power system attack risk assessment method based on multi-scene distribution
By constructing an attack success probability model based on Markov node transfer probability matrix and depth-first search algorithm, combining dynamic N-k criterion and double-layer attack-response model, the security risk assessment problem of CPPS under network attacks is solved, and the power is achieved Systematic risk assessment and defense capabilities improve.
Patent Information
- Application Number
- CN202411019202.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-29
- Publication Date
- 2024-11-22
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing technology is unable to effectively evaluate the security risks of Cyber-Physical Power System (CPPS) under cyber attacks, resulting in potential serious security threats and economic losses to the power system.
Using the attack success probability calculation model based on Markov node transfer probability matrix and depth-first search algorithm, combining dynamic N-k criterion and double-layer attack-response model, a power system attack risk assessment model under multi-scene distribution is constructed, and multi-scene attacks are proposed The expected operating cost, output adjustment and loss-load indicators are subject to quantitative evaluation.
Quantitative assessment of attack risks for CPPS at different security defense levels has been achieved, the prediction and defense capabilities of network attacks have been improved, and the safe and stable operation of the power system has been ensured.
Smart Images

Figure CN119011220A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of smart grid risk assessment, and specifically relates to a power system attack risk assessment method based on multi-scenario distribution. Background Art
[0002] Under the development trend of intelligent new power systems, a large number of electrical equipment, data acquisition equipment, computing and control equipment and decision-making units are interconnected through power grids, communication networks and information networks. The traditional power grid with electrical primary equipment as the core has evolved into a Cyber-Physical Power System (CPPS) in which the power grid and the information network are associated, integrated and interacted. Although the high degree of coupling between the information network and the power grid has greatly improved the operating efficiency of the smart grid, the continuous development of information technology will have potential negative impacts on CPPS. The security risks in the information network are likely to be introduced into the power grid, seriously affecting its reliability. In particular, malicious network attacks against CPPS by network attackers may pose a serious security threat to the power grid, triggering large-scale chain failures and causing significant economic losses.
[0003] The safety and reliability of modern power systems not only include the equipment safety of physical systems, but also focus on the information system security of CPPS.
[0004] How to effectively evaluate the security risks of CPPS under cyber attacks and strengthen the security defense of CPPS in a targeted manner before the cyber attacks come has become particularly urgent based on the evaluation data. Since the attack scenarios against CPPS in actual operation are uncertain, in order to ensure the safe and stable operation of the power system, it is imperative to study and construct a power system attack risk assessment method based on multi-scenario distribution, which has important strategic significance and practical value. Summary of the invention
[0005] In order to solve the problem that the existing technology cannot effectively evaluate the security risk of CPPS under network attacks, the present invention provides a power system attack risk assessment method based on multi-scenario distribution. First, based on the Markov node transfer probability matrix and the depth-first search algorithm, the present invention proposes a calculation model for the attack success probability of transmission line circuit breakers and substation circuit breakers. Secondly, based on the dynamic Nk criterion, a line attack interruption set under multiple scenarios is constructed. Then, based on the scenario probability distribution norm constraint and the two-layer attack-response model, a power system attack risk assessment model under multi-scenario distribution is constructed. Finally, four attack risk assessment indicators are proposed, namely, the expected operating cost index of multi-scenario attacks, the expected output adjustment index of multi-scenario attacks, the expected load loss index of multi-scenario attacks, and the comprehensive quantitative index of attack risks, to quantitatively evaluate the attack risk of CPPS under different security defense levels.
[0006] The technical solution adopted by the power system attack risk assessment method based on multi-scenario distribution of the present invention is:
[0007] A method for assessing power system attack risk based on multi-scenario distribution includes the following steps:
[0008] S1. Based on the “three layers and two networks” architecture, analyze the information-physical interaction mechanism of CPPS and the cross-domain risk propagation mechanism of cyber attacks in CPPS;
[0009] S2. According to the operation mechanism of the power cyber-physical system and the IEC 61850 standard specification, a directed attack graph targeting transmission line circuit breakers and substation circuit breakers is constructed;
[0010] S3, calculating the probability of attack entry selection, security vulnerability selection, security vulnerability attack success probability and attack success probability between logical function nodes in the network attack graph, and forming a Markov node transfer probability matrix of the network attack graph;
[0011] S4. Based on the depth-first search algorithm, the network attack graph is traversed and searched to obtain all attack paths for each circuit breaker;
[0012] S5. Calculate the attack success probability of each transmission line circuit breaker and substation circuit breaker according to the Markov node transition probability matrix obtained in S3 and all attack paths against the circuit breaker obtained in S4;
[0013] S6. Constructing a line attack interruption set under multi-scenario distribution based on the dynamic Nk criterion;
[0014] S7. Based on the scenario probability distribution norm constraint and the two-layer attack-response model, a power system attack risk assessment model under multi-scenario distribution is constructed to quantitatively assess the power system attack operation cost and load loss under multiple attack scenarios.
[0015] S8. Based on four attack risk assessment indicators, namely, the expected operating cost indicator for multi-scenario attacks, the expected output adjustment indicator for multi-scenario attacks, the expected load loss indicator for multi-scenario attacks, and the comprehensive quantitative indicator for attack risks, the attack security risks of CPPS at different security defense levels are quantitatively assessed.
[0016] A further improvement of the solution of the present invention is that in step S3
[0017] Probability of attack entry being selected:
[0018] The attackers have N E attack entrances, according to the uncertainty concept, it can be assumed that each attack entrance E i The probability of being selected is equal, that is:
[0019]
[0020] Probability of security vulnerability being selected:
[0021] The probability of selecting a security vulnerability at the attack entrance can be measured by the probability of successful attack of the security vulnerability. The probability of selecting a security vulnerability Pch(Vi) is defined as the exploitability rate P of a vulnerability. Vi The ratio of the exploitability of the vulnerability to the sum of the exploitability of all vulnerabilities that can be exploited at the attack entrance is shown in the following formula (2):
[0022]
[0023] Among them, M V The number of vulnerabilities that can be exploited for each attack entry; P Vi Vulnerability V i availability;
[0024] Probability of successful security vulnerability attack:
[0025] The attack success probability of a security vulnerability can be calculated based on the indicator parameters of the vulnerability itself, as shown in the following formula (3):
[0026]
[0027] Among them, k, α are the parameters of Pareto distribution; Vulnerability V i exposure time; Vulnerability V i Attack paths, attack complexity, and identity authentication indicator assignment;
[0028] Probability of successful attack between logical function nodes:
[0029] The attack success probability of each attack step between logical function nodes is related to the security of the CPPS information system and the attacker's attack capability. The security between logical function nodes can be quantified by the security factor λ. Every two connected logical function nodes form an attack step. The attacker's attack capability can be quantified by the attack resource c. The attack success probability of each attack step is modeled as shown in the following formula (4):
[0030]
[0031] Among them, l x is the xth attack step; x is the security factor of the xth attack step; R is the equivalent cost for the attacker to complete the attack; c is the size of the attack resource.
[0032] A further improvement of the solution of the present invention is that the information system defense decision matrix
[0033] The security factor of each attack step can be determined by a defense decision variable. Each decision variable has Y defense evaluation indicators. All defense decision variables on the information system side can form an initial defense decision matrix A, where x represents the defense decision variable of the xth attack step, and y represents the assignment of the yth evaluation indicator of the xth decision variable.
[0034]
[0035] The initial defense decision matrix is weighted and normalized using the following equations (6)-(7):
[0036]
[0037] A' x ' ,y =W y A' x,y (7)
[0038] Among them, X is the number of defense decision variables, that is, the number of all attack steps in the network attack graph; the comprehensive weight of defense evaluation index
[0039] The comprehensive weight W of the yth defense evaluation index in formula (7) is y It can be calculated based on the analytic hierarchy process-entropy weight method; according to the determining factors affecting the security performance of CPPS and the deployment of security defense measures on the information system side, the CPPS layered defense evaluation index system is divided into the target layer, the criterion layer and the indicator layer. The target layer is the security factor, the criterion layer includes 4 evaluation rules, and the indicator layer includes 12 evaluation indicators. The security factor represents the security defense level of each attack step between logical function nodes;
[0040] The analytic hierarchy process can be used to calculate the subjective weights of defense evaluation indicators. The judgment matrix B of the indicator layer is l The judgment matrix C of the criterion layer can be constructed based on the expert scoring system, as shown in the following formula (8);
[0041]
[0042] Calculate the maximum eigenvalue of each judgment matrix in turn, and perform consistency check to obtain the subjective weight matrix W from the indicator layer to the criterion layer Bl And the subjective weight matrix W from the criterion layer to the target layer C , the eigenvalue method is used to calculate the subjective weight w of each defense evaluation index y , as shown in the following formula (9);
[0043] w y =[W B1,W B2 ,W B3 ,W B4 ]W C (9)
[0044] The entropy weight method can be used to calculate the objective weight of each defense evaluation index, as shown in the following equations (10)-(14);
[0045]
[0046] e y =θ'-E y (12)
[0047]
[0048] Among them, w' y represents the objective weight of the y-th defense evaluation index; e y represents the redundancy of the y-th defense evaluation index; E y represents the entropy value of the y-th defense evaluation index; θ' is a variable adjustment parameter;
[0049] Based on the analytic hierarchy process-entropy weight method, the subjective and objective comprehensive weights W of the 12 defense evaluation indicators can be obtained. y , as shown in formula (15);
[0050]
[0051] Safety factor calculation between logical function nodes
[0052] Combining the superior and inferior solution distance method and the grey correlation analysis method to evaluate the calculation safety factor λ x ;
[0053] The defense evaluation index A' after weighted standardization of each defense decision variable x ' y The positive ideal solution of and negative ideal solution As shown in the following formula (16);
[0054]
[0055] The Euclidean distance between each defense decision variable and the ideal solution and As shown in the following formula (17);
[0056]
[0057] Grey correlation coefficient between each defense evaluation index and the ideal solution and As shown in the following formula (18);
[0058]
[0059] Among them, ρ is the resolution coefficient, and its value range is ρ∈(0,1);
[0060] Grey correlation degree of each defense decision variable and As shown in the following formula (19);
[0061]
[0062] The Euclidean distance obtained by formula (17) and the grey relational degree obtained by formula (19) are normalized as shown in formula (20);
[0063]
[0064] The approximation between each defense decision variable and the ideal solution and As shown in the following formula (21);
[0065]
[0066] Among them, f1 and f2 are the scores of the expert system for the degree of preference, and the values meet the requirement of f1+f2=1;
[0067] Based on the approximation in formula (21) and The security factor λ of each attack step between logical function nodes can be calculated x , as shown in formula (22);
[0068]
[0069] in, It is the average value of all defense evaluation indicators in the defense decision variables.
[0070] A further improvement of the solution of the present invention is that in step S3
[0071] Markov node transition probability matrix:
[0072] Substituting formula (22) into formula (4) can obtain the transition probability between nodes in the network attack graph, and the probability of selection of the comprehensive attack entrance P ch (E i ), the probability of security vulnerability being selected P ch (V i ), the probability of successful attack of security vulnerabilities And the attack success probability P between logical function nodes LNForm the Markov node transition probability matrix P pq , as shown in formula (23);
[0073]
[0074] Based on the directed adjacency matrix, the connection relationship between nodes in the network attack graph can be quantified as shown in the following formula (24);
[0075]
[0076] Where N is the total number of nodes in the network attack graph; if the information flow of the network attack graph can flow between nodes p and q, then z pq =1, otherwise z pq =0.
[0077] A further improvement of the solution of the present invention is that the success probability of the attack on the circuit breaker in S5 is:
[0078] According to the directed adjacency matrix Z of the attack graph and the depth-first search algorithm, all attack paths from the attack entry to the attack target in the network attack graph can be obtained, and the attack success probability for each circuit breaker can be calculated, as shown in the following formula (25);
[0079]
[0080] Where H is the total number of attack paths with circuit breaker n as the attack target in the network attack graph; M h is the total number of logical function nodes in the attack path h; N B is the total number of circuit breakers in the power system.
[0081] A further improvement of the solution of the present invention is that step S6 comprises:
[0082] Since the attack scenarios against CPPS in actual operation are uncertain, the line attack set under multi-scenario distribution is constructed based on the dynamic Nk criterion, as shown in the following formula (26);
[0083]
[0084] Wherein, Α(s) is the operating state set of the power system lines under attack scenario s; a l (s) is the attack running status of line l under attack scenario s, a l (s) = 0 means that line l is disconnected by the attack in attack scenario s. l (s) = 1 means that line l remains in a normal closed state under attack scenario s; R A (s) is the maximum upper limit of the number of disconnections of line l under attack scenario s.
[0085] A further improvement of the solution of the present invention is that when the power system attack risk assessment model under multi-scenario distribution is specifically constructed in step S7:
[0086] Based on the distributed blue stick idea and DC power flow model, a power system attack risk assessment model under multi-scenario distribution is established as shown in the following equations (27)-(43). It is a three-layer max-max-min structure. Based on this model, the operation cost and load shedding of the power system under multiple attack scenarios can be quantitatively quantified to evaluate the attack security risk of CPPS under different security defense levels.
[0087] The objective function is shown in Equation (27), which includes the unit output cost and load shedding penalty cost under multiple attack scenario distributions;
[0088]
[0089] Among them, p s is the probability of attack scenario s, Ω is the set of all attack scenarios; P G,g (s) is the active power output of unit g under attack scenario s, P F,l (s) is the active power flow of line l under attack scenario s, ΔP D,d (s) is the load shedding amount of load node d under attack scenario s, θ b (s) is the phase angle of node b under attack scenario s; N represents the total number of attack scenarios, N G Represents the total number of generator sets, N D Indicates the total number of load nodes;
[0090] Based on the 1-norm and ∞-norm, the probability distribution of all attack scenarios can be constrained as shown in the following equations (28)-(34);
[0091]
[0092]
[0093] Among them, Ω is the probability distribution set of all attack scenarios; θ1, θ ∞ are the attack scenario probabilities p under 1-norm and ∞-norm constraints respectively. s Deviation from initial probability The allowable deviation limit of ∞ are the uncertainty probability confidence under 1-norm and ∞-norm constraints respectively; K is the number of historical attack scenarios; m s is the linearized auxiliary variable of the 1-norm and ∞-norm constraints in equation (28);
[0094] The initial probability distribution of multiple attack scenarios can be calculated by the attack success probability for each circuit breaker in formula (25), as shown in the following formulas (35)-(36);
[0095]
[0096] Among them, P L is the line interruption probability matrix under all attack scenarios. If the line is interrupted by an attack, then P L (s,l) takes the maximum value of the success probability of the circuit breaker attack at both ends of the line, otherwise P L (s,l)=1; is the set of lines, N L is the total number of lines;
[0097] The two-layer attack-response model of the power system under attack scenario s is shown in equations (37)-(43), which is a max-min structure. The objective function is shown in equation (37), and the constraints include line attack interruption constraint (38), node power balance constraint (39), line flow constraint (40), unit output upper and lower limit constraint (41), line flow capacity constraint (42), and load node load shedding upper limit constraint (43) under attack scenario s.
[0098]
[0099]
[0100] Among them, b, l, g, and d are node b, line l, unit g, and load node d respectively; are node set, line set, unit set, and load node set respectively; K G , K L , K D They are the node-unit association matrix, node-line association matrix, and node-load association matrix respectively; X l is the reactance of line l; is the upper limit of the output of unit g; is the upper limit of the power flow capacity of line l; is the load demand of load node d; b (s), π l (s), are the dual variables of the DC power flow model constraints under attack scenario s;
[0101] Since the two-layer attack-response model of the power system in equations (37)-(43) belongs to the max-min structure and cannot be solved directly by the solver, it can be converted into a single-layer optimization problem based on the duality theory; the dual variables of the power system response model constraints based on the DC power flow model are given in equations (39)-(43), and the converted single-layer max optimization model is shown in equations (44)-(53);
[0102]
[0103]
[0104] Among them, formula (45) is the decision variable of the transformed single-layer max optimization model, including attack variables and dual variables; formula (47) is the line flow P based on the attack scenario s F,l (s) dual transformation constraint, formula (48) is the unit output P based on attack scenario s G,g (s), Equation (49) is the node phase angle θ under attack scenario s b (s), and formula (50) is the load shedding amount ΔP of the load node under the attack scenario s. D,d (s) is the dual transformation constraint; Equations (51)-(53) are the dual variable constraints of the original inequality constraints;
[0105] In addition, the constraint (49) introduces a nonlinear variable a l (s)π l (s), it can be linearized by introducing auxiliary variables, as shown in equations (54)-(57);
[0106]
[0107] Among them, t l (s), h l (s) is the linearization auxiliary variable; is the dual variable π l The minimum and maximum values of (s).
[0108] A further improvement of the solution of the present invention is that step S8 specifically includes:
[0109] Multi-scenario attack expected operating cost index R1
[0110] R1 reflects the expected operating cost of the power system under multiple attack scenarios, as shown in the following formula (58):
[0111]
[0112] Among them, P G,g , ΔPD,d They are the unit output and load shedding of the power system in a normal non-attack scenario.
[0113] Multi-scenario attack expected output adjustment index R2
[0114] R2 reflects the expected output adjustment size of the power system under multiple attack scenarios, as shown in the following formula (59):
[0115]
[0116] Multi-scenario attack expected loss load index R3
[0117] R3 reflects the expected load loss of the power system under multiple attack scenarios, as shown in the following formula (60):
[0118]
[0119] Comprehensive quantitative index of attack risk R Total
[0120] R Total It reflects the comprehensive attack risk value of the power system under multiple attack scenarios, as shown in the following formula (61):
[0121] R Total =ω1R1+ω2R2+ω3R3 (61)
[0122] Among them, ω1, ω2, and ω3 are the normalized weights of risk assessment indicators R1, R2, and R3, respectively.
[0123] The beneficial effects of the present invention are:
[0124] 1. Based on the Markov node transfer probability matrix and depth-first search algorithm, a calculation model for the attack success probability of circuit breakers is proposed to quantitatively analyze and quantify the success probability of network attacks on the information system side.
[0125] 2. Based on the dynamic Nk criterion, a line attack interruption set under multiple scenarios is constructed. Based on the scenario probability distribution norm constraint and the two-layer attack-response model, a power system attack risk assessment model under multi-scenario distribution is constructed. The economy, volatility and robustness of the power system after a successful attack are quantitatively analyzed and quantified.
[0126] 3. Four attack risk assessment indicators are proposed, namely, the expected operating cost indicator for multi-scenario attacks, the expected output adjustment indicator for multi-scenario attacks, the expected load loss indicator for multi-scenario attacks, and the comprehensive quantitative indicator for attack risks. These indicators can quantitatively assess the attack security risks of CPPS under different security defense levels. BRIEF DESCRIPTION OF THE DRAWINGS
[0127] Figure 1 It is an information-physical interaction diagram of a CPPS based on a multi-scenario distributed power system attack risk assessment method provided by the present invention;
[0128] Figure 2 It is a schematic diagram of cross-domain propagation of network attack risk in CPPS based on a multi-scenario distribution power system attack risk assessment method provided by the present invention;
[0129] Figure 3 It is a network attack graph for transmission line circuit breakers provided by the present invention, based on a multi-scenario distributed power system attack risk assessment method;
[0130] Figure 4 It is a network attack graph for substation circuit breakers provided by the present invention, based on a multi-scenario distributed power system attack risk assessment method;
[0131] Figure 5 It is a defense evaluation index weight calculation framework diagram of a power system attack risk assessment method based on multi-scenario distribution provided by the present invention;
[0132] Figure 6 It is a calculation framework diagram of the attack success probability of each circuit breaker in a power system attack risk assessment method based on multi-scenario distribution provided by the present invention;
[0133] Figure 7 It is a flow chart of power system attack risk assessment under multi-scenario distribution of a power system attack risk assessment method based on multi-scenario distribution provided by the present invention;
[0134] Figure 8 It is an attack success probability diagram of each line under different security defense schemes of a power system attack risk assessment method based on multi-scenario distribution provided by the present invention. DETAILED DESCRIPTION
[0135] In order to make the purpose, technical solution and advantages of the present invention more clear, the present invention is further described in detail below in combination with specific implementation methods and with reference to the accompanying drawings. In the following description, the description of well-known structures and technologies is omitted to avoid unnecessary confusion of the concept of the present invention.
[0136] The present invention specifically comprises the following steps:
[0137] S1. Based on the “three layers and two networks” architecture, analyze the information-physical interaction mechanism of CPPS and the cross-domain risk propagation mechanism of cyber attacks in CPPS.
[0138] The information-physical interaction mechanism of CPPS is:
[0139] CPPS uses computing, communication, and control technologies (3C technologies) to build an up-and-down closed-loop system based on situational awareness, interconnected communication, and control optimization between the information domain and the physical domain. It will provide a potential solution for the efficiency and intelligence needs of the traditional power system operation mode that integrates power generation, transformation, transmission, distribution, and power consumption. The schematic diagram of its information-physical interaction mechanism is shown in the attached figure. Figure 1 shown.
[0140] At the same time, the cross-domain risk propagation mechanism of the above-mentioned network attacks in CPPS is as follows:
[0141] The protection and control functions of CPPS can be simplified into two processes: data upload and command transmission. The measurement module collects data from the primary power equipment and transmits it to the dispatch center server. After the dispatch center analyzes and processes the data, it sends a control command to the smart terminal to control the corresponding power equipment. Network attackers can use the monitoring host, remote monitoring interface, and measurement and control merging unit as attack entrances to launch false data injection, denial of service and other attacks on the information system, causing the dispatch center to make wrong command decisions, resulting in generator shutdown or circuit breaker malfunction, and ultimately causing a chain failure of the power system. The cross-domain risk propagation diagram of network attacks in CPPS is shown in the attached figure. Figure 2 shown.
[0142] Firstly, the information equipment of CPPS is abstracted into different types of logical function nodes. Then, according to the operation mechanism of CPPS and the IEC 61850 standard, a network attack graph for line circuit breakers and transformer circuit breakers is constructed to describe the network attack process and identify the attack path. The functions of the logical function nodes are shown in Table 1. The network attack graph for transmission line circuit breakers and substation circuit breakers is shown in the attached figure. Figure 3-4 shown.
[0143] Table 1 Logical function nodes on the CPPS information system side
[0144]
[0145] S2. According to the operation mechanism of the power cyber-physical system and the IEC 61850 standard specifications, a directed attack graph targeting transmission line breakers and substation breakers is constructed.
[0146] S3. Calculate the probability of attack entry selection, security vulnerability selection, security vulnerability attack success probability and attack success probability between logical function nodes in the network attack graph to form a Markov node transfer probability matrix of the network attack graph.
[0147] The probability of the above attack entrance being selected is calculated as follows:
[0148] according to Figure 3 The network attack graph against the circuit breaker shows that the attacker has N E attack entrances, according to the uncertainty concept, it can be assumed that each attack entrance E i The probability of being selected is equal, that is,
[0149]
[0150] The probability of the above security vulnerabilities being selected is calculated as follows:
[0151] The probability of selecting a security vulnerability at the attack entrance can be measured by the probability of successful attack of the security vulnerability. ch (V i ) is defined as the exploitability of a vulnerability The ratio of the exploitability of the vulnerability to the sum of the exploitability of all vulnerabilities that can be exploited at the attack entrance is shown in the following formula (2):
[0152]
[0153] Among them, M V The number of vulnerabilities that can be exploited for each attack entry; P Vi Vulnerability V i availability.
[0154] The above security vulnerability attack success probability is calculated as follows:
[0155] The attack success probability of a security vulnerability can be calculated based on the indicator parameters of the vulnerability itself, as shown in the following formula (3):
[0156]
[0157] Among them, k, α are the parameters of Pareto distribution; Vulnerability V i exposure time; Vulnerability V i The attack paths, attack complexity, and identity authentication indicators are assigned.
[0158] The success probability of the attack between the above logical function nodes is calculated as follows:
[0159] The attack success probability of each attack step between logical function nodes is related to the security of the CPPS information system and the attacker's attack capability. The security between logical function nodes can be quantified by the security factor λ. Every two connected logical function nodes form an attack step. The attacker's attack capability can be quantified by the attack resource c. The attack success probability of each attack step is modeled as shown in the following formula (4):
[0160]
[0161] Among them, l x is the xth attack step; x is the security factor of the xth attack step; R is the equivalent cost for the attacker to complete the attack; c is the size of the attack resource.
[0162] At the same time, the above λ x The specific calculation includes the following:
[0163] (1) Information System Defense Decision Matrix
[0164] The security factor of each attack step can be determined by a defense decision variable. Each decision variable has Y defense evaluation indicators. All defense decision variables on the information system side can form an initial defense decision matrix A, where x represents the defense decision variable of the xth attack step, and y represents the assignment of the yth evaluation indicator of the xth decision variable.
[0165]
[0166] The initial defense decision matrix is weighted and normalized using the following equations (6)-(7):
[0167]
[0168] A' x ' ,y =W y A' x,y (7)
[0169] Among them, X is the number of defense decision variables, that is, the number of all attack steps in the network attack graph.
[0170] (2) Comprehensive weight of defense evaluation indicators
[0171] The comprehensive weight W of the yth defense evaluation index in formula (7) is y It can be calculated based on the analytic hierarchy process-entropy weight method; according to the determining factors affecting the security performance of CPPS and the deployment of security defense measures on the information system side, the CPPS layered defense evaluation index system is divided into the target layer, the criterion layer and the indicator layer. The target layer is the security factor, the criterion layer includes 4 evaluation rules, and the indicator layer includes 12 evaluation indicators. The security factor represents the security defense level of each attack step between logical function nodes;
[0172] The analytic hierarchy process can be used to calculate the subjective weights of defense evaluation indicators. The judgment matrix B of the indicator layer is l The judgment matrix C of the criterion layer can be constructed based on the expert scoring system, as shown in the following formula (8);
[0173]
[0174] Calculate the maximum eigenvalue of each judgment matrix in turn, and perform consistency check to obtain the subjective weight matrix W from the indicator layer to the criterion layer Bl And the subjective weight matrix W from the criterion layer to the target layer C , the eigenvalue method is used to calculate the subjective weight w of each defense evaluation index y , as shown in the following formula (9);
[0175] w y =[W B1 ,W B2 ,W B3 ,W B4 ]W C (9)
[0176] The entropy weight method can be used to calculate the objective weight of each defense evaluation index, as shown in the following equations (10)-(14);
[0177]
[0178] e y =θ'-E y (12)
[0179]
[0180] Among them, w' y represents the objective weight of the y-th defense evaluation index; e y represents the redundancy of the y-th defense evaluation index; E y represents the entropy value of the y-th defense evaluation index; θ' is a variable adjustment parameter;
[0181] Based on the analytic hierarchy process-entropy weight method, the subjective and objective comprehensive weights W of the 12 defense evaluation indicators can be obtained. y , as shown in formula (15);
[0182]
[0183] (3) Calculation of security factors between logical function nodes
[0184] The safety factor λ is calculated by combining the superior and inferior solution distance method (TOPSIS) and the grey relational analysis method (GRD) x ;
[0185] The defense evaluation index A' after weighted standardization of each defense decision variable x ' y The positive ideal solution of and negative ideal solution As shown in the following formula (16);
[0186]
[0187] The Euclidean distance between each defense decision variable and the ideal solution and As shown in the following formula (17);
[0188]
[0189] Grey correlation coefficient between each defense evaluation index and the ideal solution and As shown in the following formula (18);
[0190]
[0191] Among them, ρ is the resolution coefficient, and its value range is ρ∈(0,1);
[0192] Grey correlation degree of each defense decision variable and As shown in the following formula (19);
[0193]
[0194] The Euclidean distance obtained by formula (17) and the grey relational degree obtained by formula (19) are normalized as shown in formula (20);
[0195]
[0196] The approximation between each defense decision variable and the ideal solution and As shown in the following formula (21);
[0197]
[0198] Among them, f1 and f2 are the scores of the expert system for the degree of preference, and the values meet the requirement of f1+f2=1;
[0199] Based on the approximation in formula (21) and The security factor λ of each attack step between logical function nodes can be calculated x , as shown in formula (22);
[0200]
[0201] in, It is the average value of all defense evaluation indicators in the defense decision variables.
[0202] At the same time, the Markov node transition probability matrix is calculated in detail:
[0203] Substituting formula (22) into formula (4) can obtain the transition probability between nodes in the network attack graph, and the probability of selection of the comprehensive attack entrance P ch (E i ), the probability of security vulnerability being selected P ch (V i ), the probability of successful attack of security vulnerabilities And the attack success probability P between logical function nodes LN Form the Markov node transition probability matrix P pq , as shown in formula (23);
[0204]
[0205] Based on the directed adjacency matrix, the connection relationship between nodes in the network attack graph can be quantified as shown in the following formula (24);
[0206]
[0207] Where N is the total number of nodes in the network attack graph; if the information flow of the network attack graph can flow between nodes p and q, then z pq =1, otherwise z pq =0.
[0208] S4. Based on the depth-first search algorithm, the network attack graph is traversed and searched to obtain all attack paths for each circuit breaker.
[0209] S5. Calculate the attack success probability of each transmission line circuit breaker and substation circuit breaker according to the Markov node transfer probability matrix obtained in S3 and all attack paths for the circuit breaker obtained in S4.
[0210] The success probability of the above attack on the circuit breaker is calculated as follows:
[0211] According to the directed adjacency matrix Z of the attack graph and the depth-first search algorithm, all attack paths from the attack entrance to the attack target in the network attack graph can be obtained, and then based on P pq and attached Figure 6 , calculate the attack success probability for each circuit breaker, as shown in the following formula (25);
[0212]
[0213] Where H is the total number of attack paths with circuit breaker n as the attack target in the network attack graph; M h is the total number of logical function nodes in the attack path h; N B is the total number of circuit breakers in the power system.
[0214] S6. Construct a line attack interruption set under multi-scenario distribution based on the dynamic Nk criterion.
[0215] When the above set is specifically constructed, since the attack scenario against CPPS in actual operation is uncertain, the line attack interruption set under multi-scenario distribution is constructed based on the dynamic Nk criterion, as shown in the following formula (26);
[0216]
[0217] Wherein, Α(s) is the operating state set of the power system lines under attack scenario s; a l (s) is the attack running status of line l under attack scenario s, a l (s) = 0 means that line l is disconnected by the attack in attack scenario s. l (s) = 1 means that line l remains in a normal closed state under attack scenario s; R A (s) is the maximum upper limit of the number of disconnections of line l under attack scenario s.
[0218] S7. Based on the scenario probability distribution norm constraint and the two-layer attack-response model, a power system attack risk assessment model under multi-scenario distribution is constructed to quantitatively evaluate the power system attack operating cost and load loss under multiple attack scenarios.
[0219] When constructing the power system attack risk assessment model under the above multi-scenario distribution, the following are some of the steps:
[0220] Based on the distributed robustness idea and DC power flow model, a power system attack risk assessment model under multi-scenario distribution is established as shown in the following equations (27)-(43). It is a three-layer max-max-min structure. Based on this model, the operating cost and load shedding of the power system under multiple attack scenarios can be quantitatively quantified to evaluate the attack security risk of CPPS under different security defense levels.
[0221] The objective function is shown in Equation (27), which includes the unit output cost and load shedding penalty cost under multiple attack scenario distributions;
[0222]
[0223] Among them, p s is the probability of attack scenario s, Ω is the set of all attack scenarios; P G,g (s) is the active power output of unit g under attack scenario s, P F,l (s) is the active power flow of line l under attack scenario s, ΔP D,d (s) is the load shedding amount of load node d under attack scenario s, θ b(s) is the phase angle of node b under attack scenario s; N represents the total number of attack scenarios, N G Represents the total number of generator sets, N D Indicates the total number of load nodes.
[0224] Based on the 1-norm and ∞-norm, the probability distribution of all attack scenarios can be constrained as shown in the following equations (28)-(34);
[0225]
[0226] Among them, Ω is the probability distribution set of all attack scenarios; θ1, θ ∞ are the attack scenario probabilities p under 1-norm and ∞-norm constraints respectively. s Deviation from initial probability The allowable deviation limit of ∞ are the uncertainty probability confidence under 1-norm and ∞-norm constraints respectively; K is the number of historical attack scenarios; m s is the linearized auxiliary variable of the 1-norm and ∞-norm constraints in equation (28).
[0227] The initial probability distribution of multiple attack scenarios can be calculated by the attack success probability for each circuit breaker in formula (25), as shown in the following formulas (35)-(36);
[0228]
[0229] Among them, P L is the line interruption probability matrix under all attack scenarios. If the line is interrupted by an attack, then P L (s,l) takes the maximum value of the success probability of the circuit breaker attack at both ends of the line, otherwise P L (s,l)=1; is the set of lines, N L is the total number of lines.
[0230] The two-layer attack-response model of the power system under attack scenario s is shown in equations (37)-(43), which is a max-min structure. The objective function is shown in equation (37), and the constraints include line attack interruption constraint (38), node power balance constraint (39), line flow constraint (40), unit output upper and lower limit constraint (41), line flow capacity constraint (42), and load node load shedding upper limit constraint (43) under attack scenario s.
[0231]
[0232] Among them, b, l, g, and d are node b, line l, unit g, and load node d respectively; are node set, line set, unit set, and load node set respectively; K G , K L , K D They are the node-unit association matrix, node-line association matrix, and node-load association matrix respectively; X l is the reactance of line l; is the upper limit of the output of unit g; is the upper limit of the power flow capacity of line l; is the load demand of load node d; b (s), π l (s), are the dual variables of the DC power flow model constraints under attack scenario s.
[0233] Since the two-layer attack-response model of the power system in equations (37)-(43) belongs to the max-min structure and cannot be solved directly by the solver, it can be converted into a single-layer optimization problem based on the duality theory; the dual variables of the power system response model constraints based on the DC power flow model are given in equations (39)-(43), and the converted single-layer max optimization model is shown in equations (44)-(53);
[0234]
[0235] Among them, formula (45) is the decision variable of the transformed single-layer max optimization model, including attack variables and dual variables; formula (47) is the line flow P based on the attack scenario s F,l (s) dual transformation constraint, formula (48) is the unit output P based on attack scenario s G,g (s), Equation (49) is the node phase angle θ under attack scenario s b (s), and formula (50) is the load shedding amount ΔP of the load node under the attack scenario s. D,d (s) is the dual transformation constraint; Equations (51)-(53) are the dual variable constraints of the original inequality constraints.
[0236] In addition, the constraint (49) introduces a nonlinear variable a l (s)π l (s), it can be linearized by introducing auxiliary variables, as shown in equations (54)-(57);
[0237]
[0238] Among them, t l (s), h l (s) is the linearization auxiliary variable; is the dual variable π l The minimum and maximum values of (s).
[0239] S8. Based on four attack risk assessment indicators, namely, the expected operating cost indicator for multi-scenario attacks, the expected output adjustment indicator for multi-scenario attacks, the expected load loss indicator for multi-scenario attacks, and the comprehensive quantitative indicator for attack risks, the attack security risks of CPPS at different security defense levels are quantitatively assessed.
[0240] The expected operating cost indicator R1 of the above multi-scenario attack is calculated as follows:
[0241] The economic index R1 reflects the expected operating cost of the power system under multiple attack scenarios, as shown in the following formula (58);
[0242]
[0243] Among them, P G,g , ΔP D,d They are the unit output and load shedding of the power system in a normal non-attack scenario.
[0244] The above multi-scenario attack expected output adjustment index R2 is calculated as follows:
[0245] The volatility index R2 reflects the expected output adjustment size of the power system under multiple attack scenarios, as shown in the following formula (59);
[0246]
[0247] The above multi-scenario attack expected load loss indicator R3 is calculated as follows:
[0248] The robustness index R3 reflects the expected load loss of the power system under multiple attack scenarios, as shown in the following formula (60);
[0249]
[0250] The above attack risk comprehensive quantitative index R Total In the specific calculation:
[0251] Index R Total It reflects the comprehensive attack risk value of the power system under multiple attack scenarios, as shown in the following formula (61);
[0252] R Total =ω1R1+ω2R2+ω3R3 (61)
[0253] Among them, ω1, ω2, and ω3 are the normalized weights of risk assessment indicators R1, R2, and R3, respectively.
[0254] The flow chart of power system attack risk assessment under multi-scenario distribution is as follows: Figure 7 shown.
[0255] The present invention has been simulated and analyzed based on the IEEE RTS 79 test system, which has verified the validity of the patent of the present invention. The specific simulation results are shown below.
[0256] Taking node 1 of the network attack graph as the starting node and node 31 as the ending node, based on the directed adjacency matrix Z of the network attack graph and the depth-first search algorithm, 34 attack paths for transmission line circuit breakers are obtained, as shown in Table 2 below.
[0257] Table 2 Summary of attack paths against transmission line circuit breakers
[0258]
[0259] Taking node 1 of the network attack graph as the starting node and node 31 as the ending node, based on the directed adjacency matrix Z of the network attack graph and the depth-first search algorithm, 8 attack paths for the substation circuit breaker are obtained, as shown in Table 3 below.
[0260] Table 3 Summary of attack paths against substation circuit breakers
[0261]
[0262] The present invention conducts simulation analysis under different information system security defense scheme settings, Scheme 1: normal security defense level (A); Scheme 2: security defense level increased by 10% (1.1A); Scheme 3: security defense level increased by 20% (1.2A); Scheme 4: security defense level increased by 30% (1.3A); Scheme 5: security defense level increased by 40% (1.4A); Scheme 6: security defense level increased by 50% (1.5A). The attack success probability of each line under different security defense schemes is shown in the figure. Figure 8 shown.
[0263] Under the multi-scenario distribution setting where the line attack interruption set is (3, 4, 5), the simulation results of power system attack risks under different security defense schemes are shown in Table 4 below:
[0264] Table 4. Power system attack risk assessment results under different security defense schemes
[0265]
[0266] In the above embodiment, the present invention provides a method for assessing the attack risk of a power system based on multi-scenario distribution. Based on the Markov node transfer probability matrix and the depth-first search algorithm, the present invention proposes a calculation model for the attack success probability of the circuit breaker, and quantitatively analyzes and quantifies the success probability of the network attack on the information system side; constructs a line attack interruption set under multiple scenarios based on the dynamic Nk criterion, and constructs a power system attack risk assessment model under multi-scenario distribution based on the scenario probability distribution norm constraint and the double-layer attack-response model, and quantitatively analyzes and quantifies the economy, volatility and robustness of the power system after the attack is successful; proposes four attack risk assessment indicators, namely, a multi-scenario attack expected operating cost indicator, a multi-scenario attack expected output adjustment indicator, a multi-scenario attack expected load loss indicator, and an attack risk comprehensive quantitative indicator, which can quantitatively assess the attack security risk of CPPS under different security defense levels.
[0267] The above-described embodiments are merely descriptions of preferred implementations of the present invention, and are not intended to limit the concept and scope of the present invention. Without departing from the design concept of the present invention, various modifications and improvements made by ordinary persons in the art to the technical solution of the present invention should fall within the protection scope of the present invention, and the technical contents for which protection is sought in the present invention have been fully recorded in the claims.
Claims
1. A power system attack risk assessment method based on multi-scenario distribution, characterized by: The following steps are involved: S1. Based on the "three layers and two networks" architecture, analyze the information-physical interaction mechanism of CPPS and the cross-domain risk propagation mechanism of cyber attacks in CPPS; S2. According to the operation mechanism of the power cyber-physical system and the IEC 61850 standard specification, a directed attack graph targeting transmission line circuit breakers and substation circuit breakers is constructed; S3, calculating the probability of attack entry selection, security vulnerability selection, security vulnerability attack success probability and attack success probability between logical function nodes in the network attack graph, and forming a Markov node transfer probability matrix of the network attack graph; S4. Based on the depth-first search algorithm, the network attack graph is traversed and searched to obtain all attack paths for each circuit breaker; S5. Calculate the attack success probability of each transmission line circuit breaker and substation circuit breaker according to the Markov node transition probability matrix obtained in S3 and all attack paths against the circuit breaker obtained in S4; S6. Constructing a line attack interruption set under multi-scenario distribution based on the dynamic Nk criterion; S7. Based on the scenario probability distribution norm constraint and the two-layer attack-response model, a power system attack risk assessment model under multi-scenario distribution is constructed to quantitatively assess the power system attack operation cost and load loss under multiple attack scenarios. S8. Based on four attack risk assessment indicators, namely, the expected operating cost indicator for multi-scenario attacks, the expected output adjustment indicator for multi-scenario attacks, the expected load loss indicator for multi-scenario attacks, and the comprehensive quantitative indicator for attack risks, the attack security risks of CPPS at different security defense levels are quantitatively assessed.
2. The method for assessing power system attack risk based on multi-scenario distribution according to claim 1 is characterized in that: In step S3 Probability of attack entry being selected: The attackers have N E attack entrances, according to the uncertainty concept, it can be assumed that each attack entrance E i The probability of being selected is equal, that is: Probability of security vulnerability being selected: The probability of selecting a security vulnerability at the attack entrance can be measured by the probability of successful attack of the security vulnerability. ch (V i ) is defined as the exploitability of a vulnerability The ratio of the exploitability of the vulnerability to the sum of the exploitability of all vulnerabilities that can be exploited at the attack entrance is shown in the following formula (2): Among them, M V The number of vulnerabilities that can be exploited for each attack entry point; Vulnerability V i availability; Probability of successful security vulnerability attack: The attack success probability of a security vulnerability can be calculated based on the indicator parameters of the vulnerability itself, as shown in the following formula (3): Among them, k, α are the parameters of Pareto distribution; Vulnerability V i exposure time; Vulnerability V i Attack paths, attack complexity, and identity authentication indicator assignment; Probability of successful attack between logical function nodes: The attack success probability of each attack step between logical function nodes is related to the security of the CPPS information system and the attacker's attack capability. The security between logical function nodes can be quantified by the security factor λ. Every two connected logical function nodes form an attack step. The attacker's attack capability can be quantified by the attack resource c. The attack success probability of each attack step is modeled as shown in the following formula (4): Among them, l x is the xth attack step; x is the security factor of the xth attack step; R is the equivalent cost for the attacker to complete the attack; c is the size of the attack resource.
3. The method for assessing power system attack risk based on multi-scenario distribution according to claim 2 is characterized in that: The lambda x The specific calculation includes: Information Systems Defense Decision Matrix The security factor of each attack step can be determined by a defense decision variable. Each decision variable has Y defense evaluation indicators. All defense decision variables on the information system side can form an initial defense decision matrix A, where x represents the defense decision variable of the xth attack step, and y represents the assignment of the yth evaluation indicator of the xth decision variable. The initial defense decision matrix is weighted and normalized using the following equations (6)-(7): A' x ' ,y =W y A' x,y (7) Among them, X is the number of defense decision variables, that is, the number of all attack steps in the network attack graph; Comprehensive weight of defense evaluation indicators The comprehensive weight W of the yth defense evaluation index in formula (7) is y It can be calculated based on the analytic hierarchy process-entropy weight method; according to the determining factors affecting the security performance of CPPS and the deployment of security defense measures on the information system side, the CPPS layered defense evaluation index system is divided into the target layer, the criterion layer and the indicator layer. The target layer is the security factor, the criterion layer includes 4 evaluation rules, and the indicator layer includes 12 evaluation indicators. The security factor represents the security defense level of each attack step between logical function nodes; The analytic hierarchy process can be used to calculate the subjective weights of defense evaluation indicators. The judgment matrix B of the indicator layer is l The judgment matrix C of the criterion layer can be constructed based on the expert scoring system, as shown in the following formula (8); Calculate the maximum eigenvalue of each judgment matrix in turn, and perform consistency check to obtain the subjective weight matrix W from the indicator layer to the criterion layer Bl And the subjective weight matrix W from the criterion layer to the target layer C , the eigenvalue method is used to calculate the subjective weight w of each defense evaluation index y , as shown in the following formula (9); In y =[W B1 ,IN B2 ,IN B3 ,IN B4 ]IN C (9) The entropy weight method can be used to calculate the objective weight of each defense evaluation index, as shown in the following equations (10)-(14); And y =θ'-E y (12) Among them, w' y represents the objective weight of the y-th defense evaluation index; e y represents the redundancy of the y-th defense evaluation index; E y represents the entropy value of the y-th defense evaluation index; θ' is a variable adjustment parameter; Based on the analytic hierarchy process-entropy weight method, the subjective and objective comprehensive weights W of the 12 defense evaluation indicators can be obtained. y , as shown in formula (15); Safety factor calculation between logical function nodes Combining the superior and inferior solution distance method and the grey correlation analysis method to evaluate the calculation safety factor λ x ; The defense evaluation index A' after weighted standardization of each defense decision variable x ' y The positive ideal solution of and negative ideal solution As shown in the following formula (16); The Euclidean distance between each defense decision variable and the ideal solution and As shown in the following formula (17); Grey correlation coefficient between each defense evaluation index and the ideal solution and As shown in the following formula (18); Among them, ρ is the resolution coefficient, and its value range is ρ∈(0,1); Grey correlation degree of each defense decision variable and As shown in the following formula (19); The Euclidean distance obtained by formula (17) and the grey relational degree obtained by formula (19) are normalized as shown in formula (20); The approximation between each defense decision variable and the ideal solution and As shown in the following formula (21); Among them, f1 and f2 are the scores of the expert system for the degree of preference, and the values meet the requirement of f1+f2=1; Based on the approximation in formula (21) and The security factor λ of each attack step between logical function nodes can be calculated x , as shown in formula (22); in, It is the average value of all defense evaluation indicators in the defense decision variables.
4. The method for assessing power system attack risk based on multi-scenario distribution according to claim 3 is characterized in that: In step S3 Markov node transition probability matrix: Substituting formula (22) into formula (4) can obtain the transition probability between nodes in the network attack graph, and the probability of selection of the comprehensive attack entrance P ch (E i ), the probability of security vulnerability being selected P ch (V i ), the probability of successful attack of security vulnerabilities And the attack success probability P between logical function nodes LN Form the Markov node transition probability matrix P pq , as shown in formula (23); Based on the directed adjacency matrix, the connection relationship between nodes in the network attack graph can be quantified as shown in the following formula (24); Where N is the total number of nodes in the network attack graph; if the information flow of the network attack graph can flow between nodes p and q, then z pq =1, otherwise z pq =0.
5. The method for assessing power system attack risk based on multi-scenario distribution according to claim 1 is characterized in that: The success probability of the attack on the circuit breaker in S5 is: According to the directed adjacency matrix Z of the attack graph and the depth-first search algorithm, all attack paths from the attack entry to the attack target in the network attack graph can be obtained, and the attack success probability for each circuit breaker can be calculated, as shown in the following formula (25); Where H is the total number of attack paths with circuit breaker n as the attack target in the network attack graph; M h is the total number of logical function nodes in the attack path h; N B is the total number of circuit breakers in the power system.
6. The method for assessing power system attack risk based on multi-scenario distribution according to claim 1 is characterized in that: The step S6 comprises: Since the attack scenarios against CPPS in actual operation are uncertain, the line attack interruption set under multi-scenario distribution is constructed based on the dynamic Nk criterion, as shown in the following formula (26); Wherein, Α(s) is the operating state set of the power system lines under attack scenario s; a l (s) is the attack running status of line l under attack scenario s, a l (s) = 0 means that line l is disconnected by the attack in attack scenario s. l (s) = 1 means that line l remains in a normal closed state under attack scenario s; R A (s) is the maximum upper limit of the number of disconnections of line l under attack scenario s.
7. The method for assessing power system attack risk based on multi-scenario distribution according to claim 1 is characterized by: When the power system attack risk assessment model under multi-scenario distribution is specifically constructed in step S7: Based on the distributed blue stick idea and DC power flow model, a power system attack risk assessment model under multi-scenario distribution is established as shown in the following equations (27)-(43). It is a three-layer max-max-min structure. Based on this model, the operation cost and load shedding of the power system under multiple attack scenarios can be quantitatively quantified to evaluate the attack security risk of CPPS under different security defense levels. The objective function is shown in Equation (27), which includes the unit output cost and load shedding penalty cost under multiple attack scenario distributions; Among them, p s is the probability of attack scenario s, Ω is the set of all attack scenarios; P G,g (s) is the active power output of unit g under attack scenario s, P F,l (s) is the active power flow of line l under attack scenario s, ΔP D,d (s) is the load shedding amount of load node d under attack scenario s, θ b (s) is the phase angle of node b under attack scenario s; N represents the total number of attack scenarios, N G Represents the total number of generator sets, N D Indicates the total number of load nodes; Based on the 1-norm and ∞-norm, the probability distribution of all attack scenarios can be constrained as shown in the following equations (28)-(34); Among them, Ω is the probability distribution set of all attack scenarios; θ1, θ ∞ are the attack scenario probabilities p under 1-norm and ∞-norm constraints respectively. s Deviation from initial probability The allowable deviation limit of ∞ are the uncertainty probability confidence under 1-norm and ∞-norm constraints respectively; K is the number of historical attack scenarios; m s is the linearized auxiliary variable of the 1-norm and ∞-norm constraints in equation (28); The initial probability distribution of multiple attack scenarios can be calculated by the attack success probability for each circuit breaker in formula (25), as shown in the following formulas (35)-(36); Among them, P L is the line interruption probability matrix under all attack scenarios. If the line is interrupted by an attack, then P L (s,l) takes the maximum value of the success probability of the circuit breaker attack at both ends of the line, otherwise P L (s,l)=1; is the set of lines, N L is the total number of lines; The two-layer attack-response model of the power system under attack scenario s is shown in equations (37)-(43), which is a max-min structure. The objective function is shown in equation (37), and the constraints include line attack interruption constraint (38), node power balance constraint (39), line flow constraint (40), unit output upper and lower limit constraint (41), line flow capacity constraint (42), and load node load shedding upper limit constraint (43) under attack scenario s. Among them, b, l, g, and d are node b, line l, unit g, and load node d respectively; are node set, line set, unit set, and load node set respectively; K G , K L , K D They are the node-unit association matrix, node-line association matrix, and node-load association matrix respectively; X l is the reactance of line l; is the upper limit of the output of unit g; is the upper limit of the power flow capacity of line l; is the load demand of load node d; b (s), π l (s), are the dual variables constrained by the DC power flow model under attack scenario s; Since the two-layer attack-response model of the power system in equations (37)-(43) belongs to the max-min structure and cannot be solved directly by the solver, it can be converted into a single-layer optimization problem based on the duality theory; the dual variables of the power system response model constraints based on the DC power flow model are given in equations (39)-(43), and the converted single-layer max optimization model is shown in equations (44)-(53); Among them, formula (45) is the decision variable of the transformed single-layer max optimization model, including attack variables and dual variables; formula (47) is the line flow P based on the attack scenario s F,l (s) dual transformation constraint, formula (48) is the unit output P based on attack scenario s G,g (s), Equation (49) is the node phase angle θ under attack scenario s b (s), and formula (50) is the load shedding amount ΔP of the load node under the attack scenario s. D,d (s) is the dual transformation constraint; Equations (51)-(53) are the dual variable constraints of the original inequality constraints; In addition, the constraint (49) introduces a nonlinear variable a l (s)π l (s), it can be linearized by introducing auxiliary variables, as shown in equations (54)-(57); Among them, t l (s), h l (s) is the linearization auxiliary variable; is the dual variable π l The minimum and maximum values of (s).
8. The method for assessing power system attack risk based on multi-scenario distribution according to claim 1 is characterized by: The step S8 specifically includes: Multi-scenario attack expected operating cost index R1 R1 reflects the expected operating cost of the power system under multiple attack scenarios, as shown in the following formula (58): Among them, P G,g , ΔP D,d They are the unit output and load shedding of the power system in a normal non-attack scenario. Multi-scenario attack expected output adjustment index R2 R2 reflects the expected output adjustment size of the power system under multiple attack scenarios, as shown in the following formula (59): Multi-scenario attack expected loss load index R3 R3 reflects the expected load loss of the power system under multiple attack scenarios, as shown in the following formula (60): Comprehensive quantitative index of attack risk R Total R Total It reflects the comprehensive attack risk value of the power system under multiple attack scenarios, as shown in the following formula (61): R Total =ω1R1+ω2R2+ω3R3 (61) Among them, ω1, ω2, and ω3 are the normalized weights of risk assessment indicators R1, R2, and R3, respectively.
Citation Information
Patent Citations
Power system risk assessment method considering transformer substation network attack success probability
CN115049270A
Power distribution information physical system security risk assessment method considering network attack vulnerabilities
CN115801460A
Cited By
Power system network attack consequence evaluation method considering switch modularization
CN120337742A
A method for assessing consequences of cyber attacks on power system networks considering switching modularization
CN120337742B
