CN111641595B - Power network security risk assessment method and system - Google Patents
Power network security risk assessment method and system Download PDFInfo
- Publication number
- CN111641595B CN111641595B CN202010392394.5A CN202010392394A CN111641595B CN 111641595 B CN111641595 B CN 111641595B CN 202010392394 A CN202010392394 A CN 202010392394A CN 111641595 B CN111641595 B CN 111641595B
- Authority
- CN
- China
- Prior art keywords
- power network
- security risk
- constraint
- network security
- risk assessment
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 76
- 238000012502 risk assessment Methods 0.000 title claims abstract description 40
- 238000005457 optimization Methods 0.000 claims abstract description 25
- 230000009194 climbing Effects 0.000 claims abstract description 24
- 230000008602 contraction Effects 0.000 claims abstract description 8
- 239000011159 matrix material Substances 0.000 claims description 25
- 239000013598 vector Substances 0.000 claims description 20
- 230000009977 dual effect Effects 0.000 claims description 11
- 108010074506 Transfer Factor Proteins 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 2
- 230000007547 defect Effects 0.000 abstract description 6
- 238000013459 approach Methods 0.000 abstract description 4
- 238000004422 calculation algorithm Methods 0.000 abstract description 4
- 238000012545 processing Methods 0.000 abstract description 3
- 238000004364 calculation method Methods 0.000 description 34
- 238000012360 testing method Methods 0.000 description 12
- 230000007123 defense Effects 0.000 description 4
- 230000010355 oscillation Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 238000013210 evaluation model Methods 0.000 description 3
- 238000004088 simulation Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012821 model calculation Methods 0.000 description 1
- 239000000758 substrate Substances 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L63/00—Network architectures or network communication protocols for network security
- H04L63/14—Network architectures or network communication protocols for network security for detecting or protecting against malicious traffic
- H04L63/1408—Network architectures or network communication protocols for network security for detecting or protecting against malicious traffic by monitoring network traffic
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/40—Business processes related to the transportation industry
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L63/00—Network architectures or network communication protocols for network security
- H04L63/14—Network architectures or network communication protocols for network security for detecting or protecting against malicious traffic
- H04L63/1441—Countermeasures against malicious traffic
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Computer Security & Cryptography (AREA)
- Entrepreneurship & Innovation (AREA)
- Theoretical Computer Science (AREA)
- Marketing (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Educational Administration (AREA)
- Development Economics (AREA)
- Health & Medical Sciences (AREA)
- Game Theory and Decision Science (AREA)
- Quality & Reliability (AREA)
- Computer Hardware Design (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Operations Research (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses a method and a system for evaluating the security risk of a power network, wherein a double-layer planning model for evaluating the security risk of the power network is established; obtaining an approximate optimal solution of the power network security risk assessment double-layer planning model by using a constraint and delimitation contraction method; and taking the approximate solution as an initial solution of a hill climbing method, and iteratively obtaining a global optimization risk value of the power network security risk assessment double-layer planning model. The invention adopts the idea of iterative solution of upper-layer planning and lower-layer planning to reduce the computational complexity, has less iteration times, can quickly obtain the approximate solution of the model, and overcomes the defect of low computational efficiency of processing a large-scale power system by using the traditional method. The method limits a plurality of local optimal solutions of the double-layer planning model outside the range of solving the search domain by adopting a constraint and delimitation contraction means, provides an effective initial value for the iterative solution process, enables the final iterative result to approach or be equal to the global optimal solution, and effectively overcomes the defect of low solution precision of the traditional iterative solution algorithm.
Description
Technical Field
The invention relates to the technical field of information physical security of power systems, in particular to a power network security risk assessment method and system.
Background
In recent years, in order to increase the level of intelligence of an electric power system, information technology and communication technology have been widely used in an electric power network. The high integration of information technology inevitably introduces corresponding cyber-security risks, making the power system measurement data vulnerable to tampering attacks. The accuracy of the measured data is of great importance to the scheduling safety of the power system, and an attacker can destroy the integrity of the collected measured data by injecting malicious data and mislead a scheduler to make wrong decisions so as to cause safety accidents such as line disconnection, load shedding, cascading failure and the like.
Therefore, a power network security risk assessment model facing malicious data attack is urgently needed to be developed from the defense perspective, potential risks are pre-judged, and a reference basis is provided for defense strategies. Mathematically, the power system security risk assessment under network attack can be expressed as a two-tier planning problem.
The most common approach to solving the two-layer planning model is to introduce the Karush-Kuhn-Tucker (KKT) condition. Although the global optimal solution of the model can be obtained by the calculation method based on the KKT, a large amount of additional integer variables and constraints are required to be introduced in the solving process, so that the calculation efficiency is extremely low, and the method is difficult to process a large-scale power system. However, the existing other heuristic methods cannot guarantee the calculation accuracy, for example, a hill climbing method is usually easy to fall into local optimal oscillation, although the calculation efficiency is high, the difference between the obtained final solution and the global optimal solution is possibly large, thereby causing system risk misjudgment. How to rapidly calculate the global optimal network security risk value of the power system is a problem to be solved urgently at present.
Disclosure of Invention
The technical problem to be solved by the invention is to provide a power network security risk method and system aiming at the defects of the prior art, overcome the local optimal solution oscillation phenomenon existing in the prior art, and improve the calculation accuracy and the calculation efficiency of the risk value.
In order to solve the technical problems, the technical scheme adopted by the invention is as follows: a power network security risk assessment method comprises the following steps:
1) establishing a power network security risk assessment double-layer planning model;
2) obtaining an approximate optimal solution of the power network security risk assessment double-layer planning model by using a constraint and delimitation contraction method;
3) and taking the approximate solution as an initial solution of a hill climbing method, and iteratively obtaining a global optimization risk value of the power network security risk assessment double-layer planning model.
The method overcomes the local optimal oscillation solving phenomenon existing in the traditional method, achieves the aim of finding the global optimization risk value of the power network security risk assessment model, namely optimizing the risk value, in a short time, and improves the calculation accuracy and the calculation efficiency of the risk value.
In the invention, the expression of the electric power network security risk assessment double-layer planning model is as follows:
maxycTx;
s.t.
Ey≤g;
minxcTx;
s.t.
Ax≤b-By(λ);
wherein, cTx represents a risk indicator of the power network; y represents an attack vector; x represents the power network operating state; a and B are coefficient matrixes corresponding to variables x and y; λ represents the lagrange multiplier; b is a constant matrix of the lower layer constraint; e is a coefficient matrix of the variable y in the upper layer constraint; g is a constant matrix of upper layer constraints.
The invention establishes the power network security risk assessment model comprehensively considering the angles of attackers and defenders, abstracts the model into a Max-min double-layer planning model, can effectively mine the potential network security risk of the system, and has important significance for making a power network security defense strategy.
The specific implementation process of the step 2) comprises the following steps:
i. for the power network security risk assessment double-layer planning model, an initial value of an upper-layer malicious attack vector is givenAnd initializing tight constraint sets
For a given upperLayer attack vectorObtaining the optimal operation state x of the system corresponding to the bottom optimization problem P3*Lagrange multiplier ofThe corresponding objective function value is K;
for the current upper layer attack vectorUsing the optimal operating state x of the system*Calculating the load level ε of each constraint i based on the constraint matrixi;
ifDefining the ith constraint as a tight constraint and storing the tight constraint in S, otherwise, discarding the tight constraint;is a set constant;
v. updating constraints stored in S for the current iteration processAnd returning to the step 2) until the objective function value K is not changed any more, at the momentThe optimal value of the variable y is the approximate optimal solution.
The bottom optimization problem P3 of the present invention is expressed as:
K=minxcTx;
s.t.
wherein, cTx is the system windIndex of risk, cTIs a correlation coefficient; b is a constant matrix of the lower layer constraint; a and B are coefficient matrixes of lower layer constraint variables x and y respectively.
In the step iv, the step (iii),wherein,Ai、bi、Bithe ith row of coefficient matrices a, B and B, respectively. The step has the technical advantages that a small amount of tight constraints can be screened out, the processing of the optimization problem P4 on all constraints is avoided, compared with the prior art, the number of constraints of the optimization problem is obviously reduced, and the calculation efficiency is improved.
max∑i∈S(Bi·Δy)·ωi;
s.t.
Wherein, BiΔ y represents the load increment of constraint i due to the injected malicious data Δ y, ωiA weight factor for the current constraint i load increment index normalization;expression obtained from P3A lagrange multiplier; g represents a constant matrix in a constraint that limits the attack vector; solving for P4 to get Δ y, then updating the next iterationThe value of (c):and ← denotes assignment. Compared with the prior art, the variableThe update of (1) does not need to solve the dual problem of the lower-layer optimization problem P3, does not need the lower-layer optimization problem to satisfy the condition of linear programming, and can be nonlinear and contain discrete variables. Therefore, the method has wider application range.
The specific implementation process of the step 3) of the invention comprises the following steps:
subjecting the product of step v)Endowing a hill climbing method as an iteration initial value; for the iteration initial value, solving the bottom layer optimization problem to obtain an objective function value K and a dual variable
Will duality variableUpdates in dual form P5 substituted into the underlying planAnd obtaining a corresponding objective function value f;
to be updatedSubstituting into step vi), repeating steps vi) -viii) until the two objective function values K and f are not increased any more, or the absolute difference between K and fAnd when the value is smaller than the set threshold eta, the system global optimization risk value is considered to be obtained, and the iteration is stopped.
In step vii, the expression of the dual form P5 of the bottom layer plan is:
s.t.
Ey≤g。
compared with the prior art, the steps i to v provide an optimal initial value for the iteration of the dual problem, so that the final iteration result can approach or be equal to the global optimal solution, the defect of low calculation precision of the traditional method is overcome, and the calculation precision of the optimal risk value is improved.
x=[PNG×1,JND×1,PLNL×1]T;y=ΔDND×1;cT=[01×NG,11×ND,01×NL];
N, NG, ND and NL are the number of nodes of the power system, the number of generators, the number of loads and the number of lines respectively; p is the output of the generator and has the dimension of NG multiplied by 1 and DaΔ D and J are load data, injected dummy data and load shedding caused by attack, respectively, and the dimensions are ND × 1; PL is line power flow, and the dimension of PL is NL multiplied by 1; SF is a transfer factor matrix with the dimension of NL multiplied by N; KP, KD and KL respectively represent a node-generator, a node-load and a node-line incidence matrix, and the dimensionalities of the incidence matrixes are NXNG, NXND and NXNL respectively; pmin、PmaxRespectively, the minimum value and the maximum value of P; PLmin、PLmaxRespectively, the minimum and maximum values of PL.
The invention also provides a power network security risk assessment system, which comprises computer equipment; the computer device is programmed or configured to perform the steps of the above-described power network security risk assessment method; or a storage medium of the computer device stores a program for executing the above power network security risk assessment method.
Compared with the prior art, the invention has the beneficial effects that:
(1) aiming at the operation risk of the power system under the threat of network attack, the invention establishes a power network security risk assessment model comprehensively considering the angles of attackers and defenders, and abstracts the model into a Max-min double-layer planning model. The evaluation model can effectively mine potential network security risks of the system, and has important significance for making a security defense strategy of the power network.
(2) The calculation speed is fast: the invention adopts the idea of iterative computation of upper-layer planning and lower-layer planning to reduce the computation complexity, has less iteration times, can quickly obtain the global optimization risk value, and overcomes the defect of low computation efficiency of the traditional method for processing a large-scale power system.
(3) The calculation precision is high: the method limits a plurality of local optimal solutions of the double-layer planning model outside the range of solving the search domain by adopting a constraint and bound contraction method, provides an effective initial value for the iterative computation process, enables the final iterative result to approach or be equal to the global optimization risk value, and effectively overcomes the defect of low computation precision of the traditional method.
Drawings
Fig. 1 is a power network security risk double-layer assessment model established by the invention.
FIG. 2 is a basic schematic diagram of the evaluation model calculation method according to the present invention.
Fig. 3 is a flow chart of an implementation of the present invention.
Detailed Description
The invention provides a double-layer model for power network security risk assessment and a calculation method thereof. The evaluation model calculation method is divided into two stages, and firstly, in the first stage, an approximate optimal solution of the model is obtained through a constraint and delimitation contraction algorithm. And in the second stage, the approximate optimal solution obtained in the first stage is used as an initial value and is given to a hill climbing method for iterative computation of the network security risk value of the system.
Mathematically, the general expression of the risk assessment two-tier optimization model is as follows:
maxycTx (1)
s.t.
Ey≤g (2)
minxcTx (3)
s.t.
Ax≤b-By(λ) (4)
wherein the value of the objective function cTx represents the system risk, the upper variable y represents the injected attack vector, and the lower variable x represents the system running state. Constraint (2) represents the feasible domain of variable y and constraint (4) represents the set of all equality and inequality power flow constraints in the bottom layer. E is the coefficient matrix of the variable y in the upper layer constraint, and g is the constant matrix of the upper layer constraint. A and B are coefficient matrixes of lower-layer constraint variables x and y respectively, and B is a constant matrix of lower-layer constraint. λ represents the corresponding lagrange multiplier of all the equations and inequalities at the bottom. For different system risk indicators cTx, their corresponding coefficients and constant matrices a, B, E, g, B will also be different.
Without loss of generality, the system load shedding under the attack of malicious data is used as a risk assessment index, and the double-layer optimization models (1) to (4) can be expanded into specific forms of (5) to (13):
maxΔD1TJ (5)
s.t.
1TΔD=0 (6)
-τDa≤ΔD≤τDa (7)
minP,J,PL1TJ (8)
s.t.
1TP=1TDa-1TJ (9)
PL=SF·(KP·P-KD(Da-ΔD-J)) (10)
Pmin≤P≤Pmax (11)
0≤J≤Da (12)
PLmin≤PL≤PLmax (13)
from an attacker perspective, the upper layer planning target is to maximize the load shedding amount, and from a defender perspective, the lower layer planning target is to minimize the load shedding amount. The control variable for the upper level plan is Δ D, representing the spurious data injected into the load, DaThe load data is the load data after being tampered. The injected dummy data vector must satisfy two constraints: 1) the sum of all elements of the attack vector must be zero (6); 2) all elements of the attack vector must be limited within a certain range (7), wherein the value of tau is 0-1. The bottom layer problem control variables are the output P of the generator, the shear load J caused by the excessive output of the generator and the line tide PL. The constraint (9) is a power balance equation, and the constraint (10) calculates the line load flow. Constraints (11) - (13) set the upper and lower limits of generator output P, shear load J and line current PL, respectively. SF is a transfer factor matrix, KP, KD and KL are respectively a node-generator, a node-load and a node-line incidence matrix.
Further: assuming that the number of nodes of the tested system is N, the dimensionality of the output P of the generator is NG multiplied by 1, and the load data DaThe dimension of the injected dummy data Δ D and the tangential load J is ND × 1, and the dimension of the line power flow PL is NL × 1. Then the dimensions of SF are NL × N, KP, KD and KL are nxng, nxnd, nxnl, respectively, and the variables, coefficients and constant matrices in the general form can be expressed as:
x=[PNG×1,JND×1,PNL×1]T y=ΔDND×1 cT=[01×NG,11×ND,01×NL]
FIG. 2 shows the principle of solving the approximate network security risk value by using the constrained bounding contraction algorithm, in which the closed loop represents the union of the feasible points with equal objective function valuesClosed-loop curves, i.e. identical characters (alpha) in the figure1,α2,α3,α4) The closed loop curves represented have equal objective function values and increase from the outside to the inside. Therefore, in fig. 2, it is assumed that the central point of the closed curve located at the upper left in the diagram is a local optimal solution, and the central point of the closed curve located at the lower right is a global optimal solution. The left graph shows that for a random initial value, iteration is performed by using a hill climbing method, and the final result is trapped in the marked local optimal solution with a high probability. The optimal solution to the underlying problem is x*There are some constraints that belong to bounded tight constraints. For this part of the constraint, if we update the initial value y to further reduce the value of the right term b-By of the inequality (b is a constant matrix of the underlying planning constraint), then the optimal solution x*These inequality constraints are violated, i.e. the optimal solution x for the new y*Becomes impractical. In other words, the new y is equivalent to adding several constraints to compress the solution search domain of the underlying optimization problem, so that the current local optimal solution is limited to be outside the solution search domain (the area enclosed by the dotted line), and the network risk value c is obtainedTx increases.
Further, to ensure that in the first stage, a new value is obtained for each iterationThe objective function is incremented, making the bottom of the two-layer planning model P1:
P1:min cTx (14)
s.t.
Ax≤b-Fy(λ) (15)
further, for a given upper variable y(k)Let the optimal operating state of the system solved by P1 be x*According to the dual theory
λT(b-Fy(k))=cTx* (16)
Let K be the objective function value (system risk value) of P1 in the current iteration, i.e., K ═ cTx*. Then it can be ensured that the value of the objective function is per time defined by introducing a constraint (9) to form P2The sub-iterations are incremented.
P2:min cTx (17)
s.t. constraint (16)
λT(b-Fy)≥K (18)
The invention provides a calculation method for double-layer planning of power network security risk assessment, which is specifically carried out according to the following steps:
step 1: in the execution stage one, for the double-layer planning model for evaluating the security risk of the power network, the initial value of an upper-layer attack vector is givenAnd initializing tight constraint sets
Step 2: for a given upper layer attack vectorSolving the bottom layer optimization problem P3, and recording the system optimization operation state as x*Lagrange multiplier ofThe corresponding objective function value is K.
Further, the P3 model is as follows
P3:K=minxcTx (19)
s.t.
cTx is a system risk indicator, cTIs the correlation coefficient.
And step 3: for the current upper variableUsing the optimal operating state x of the system*Then, each constraint i is calculated from a constraint matrix (20)Load level epsiloni,
Wherein u isiAnd viRespectively for optimizing the operating state x at the current system*Next, the absolute value of the left term and the right term in each constraint i in equation (20);
Ai,biand BiThe ith row of coefficient matrices a, B and B, respectively.
And 4, step 4: and determining a tight constraint set S according to the calculation result of the step 3. If it is not(For a set parameter, say around 0.9), then the ith constraint is defined as a tight constraint stored in S, otherwise the constraint is dropped. Namely:
S=S∪i (23)
the left side of equation (23) is the updated tight constraint set and the right side is the pre-update tight constraint set.
And 5: for the constraints stored in S during the current iteration, the underlying model P4 is solved to update
P4:max∑i∈S(Bi·Δy)·ωi (24)
s.t.
Further, BiΔ y represents the load increment of constraint i due to the injected attack vector Δ y, ωiAnd (4) a weight factor for normalizing the current constraint i load increment index.The lagrange multiplier obtained from P3 is shown. g denotes a constant matrix in the constraint that limits the attack vector. Constraints (26) ensure updated attack vectorsStill remaining within the set range. Constraints (27) ensure that the objective function produces an increment after each iteration. Get Δ y by solving for P4, then update the next iterationThe value of (c):
left side of equation (28)For updated values, right sideFor the pre-update value, equation (28) is expressed byTo compress the solution search domain. In the embodiment of the present invention, the first and second substrates,corresponding to the attack vector deltad.
NovelEquivalently, compressing the solution search domain of the underlying optimization problem by adding the constraint in the iteration.
Step 6: in finding out newThen, returning to step 2, and continuously performing loop iteration until the objective function value K is not changed any more, so that the user can think that the objective function value K is changed at the momentIs the best value of the variable y.
And 7: performing the second step, and mixing the result obtained in the step 6And endowing the hill climbing method as an iteration initial value. Initial value for given upper variableSolving P3 to obtain the target function value K and dual variable
And 8: then the obtained dual variables are usedUpdates in dual form P5 substituted into the underlying planAnd a corresponding objective function value f is obtained.
s.t.
Ey≤g (30)
And step 9: the iteration is continuously updated until when the two objective function values (K and f) are no longer increased, or the difference between K and f is smaller than a set threshold η (e.g., | K-f | < η), it is considered that the global optimization risk value has been obtained and the iteration is stopped.
Examples
In the invention, an IEEE RTS-24 node system is used for testing the established power network security risk assessment model, and the load shedding is selected as a system network security risk quantitative index without loss of generality. And aiming at an IEEE RTS-24 node system, comparing the obtained results by adopting a calculation method based on KKT, a hill climbing method and a method provided by the invention. Comparative results are given in the attached tables 1 to 4.
Table 1: test result based on KKT condition calculation method
Table 2: test result based on hill climbing method
Table 3: test results based on the method proposed by the invention
Aiming at the load level in a test system, without loss of generality, 1.0, 1.1, 1.2, 1.3, 1.4 and 1.5 times of reference load is selected as the load level, and the final load shedding amount and the solving time calculated by a model are compared.
Table 1 gives the simulation results of the calculation method based on the KKT condition. The second column gives the global optimal solution for the two-level planning model, and it can be seen that as the load level increases, the amount of shear load increases. In addition, we can see that the calculation method based on the KKT condition is computationally inefficient. For example, a 24-node system with a load level of 1.5 times requires more than 6 minutes of computation time to obtain the final solution. For large-scale power networks, such as IEEE 118 node systems, testing has shown that a typical personal computer takes more than ten hours to obtain a final solution.
TABLE 1 test results based on KKT Condition calculation method
Further, table 2 shows the simulation results based on the hill climbing method. For an IEEE RTS-24 node system, each sample can obtain a final solution only through one iteration, and the result reflects the rapid calculation advantage of the hill climbing method. The hill climbing method only uses less than 0.1 second to obtain a local optimal solution, and compared with a calculation method based on a KKT condition, the calculation time is shortened by thousands of times. Therefore, the characteristic of high computational efficiency of the hill climbing method is very suitable for being applied to risk assessment of a large-scale power system. However, the local optimal solution obtained by the hill climbing method has a non-negligible gap from the global optimal solution of the calculation method based on the KKT condition. For example, when the load levels are 1.1 and 1.2, the load is zero in hill climbing, and the global optimal solution determined by the calculation method based on the KKT condition is 40.9MW and 182.9MW, respectively. These data with large errors will affect the safety decisions of the dispatcher.
TABLE 2 mountain climbing based test results
For the evaluation method proposed by the present invention, first, the threshold for tight constraint i is set to 90%, i.e. the threshold for tight constraint i is set toThe simulation results are shown in table 3. Compared with a hill climbing method, most results of the method provided by the invention are closer to the optimal solution. For example, when the load level is 1.1, the hill climbing method cannot be effectiveBut the solution based on the proposed method of the present invention is 34.61 MW. Taking the load level of 1.2 as an example, the method provided by the invention can obtain a global optimal solution equivalent to a calculation method based on the KKT condition, and only takes 0.124 second. Experimental results show that the method integrates the advantages of the KKT calculation method and the hill climbing method, is superior to the KKT calculation method in solving efficiency, and is superior to the hill climbing method in model solution optimality.
TABLE 3 test results based on the proposed method of the present invention
It is to be noted that due to the parametersImproper values of (a) may cause K to fall into the same locally optimal solution as the hill-climbing method. From the last two samples of table 3, we can see that the iterative result of finding the initial value (i.e., the value of K) is equal to the final solution of the hill-climbing method. But we can further optimizeFor example, the test results of different parameters (0.85-0.89) for the 1.4 and 1.5 times load level samples are shown in table 4. It can be seen that when the parameters are chosen to be 0.87 and 0.86, respectively, both samples can find a globally optimal solution. Although the time consumed is also increased, the increased time consumption is negligible compared to the calculation speed of the calculation method based on the KKT condition. On the other hand, although there are also individual samples that do not yield a globally optimal solution, such as samples with load levels of 1.1 and 1.3 times, respectively, these two solutions are very similar to the optimal solution. Therefore, compared with a hill climbing method, the evaluation method provided by the invention has the advantage that the model calculation precision is obviously improved.
In summary, the invention establishes a double-layer planning model for power network security risk assessment aiming at the system network security risk of the power grid after being subjected to network attack, and provides a new rapid calculation method for power network security risk assessment to assess the power network security risk. The method is divided into two stages, a constraint and delimitation contraction algorithm is used for obtaining a model approximate optimal solution in the first stage, then the approximate solution is used as an initial solution of a hill climbing method in the second stage, and the power network safety risk value is obtained through gradual iteration. The invention overcomes the local optimal oscillation solving phenomenon existing in the traditional method and realizes the finding of the global optimal network risk value of the power system in a short time.
Claims (7)
1. A power network security risk assessment method is characterized by comprising the following steps:
1) establishing a power network security risk assessment double-layer planning model;
2) obtaining an approximate optimal solution of the power network security risk assessment double-layer planning model by using a constraint and delimitation contraction method;
the specific implementation process of the step 2) comprises the following steps:
i. for the power network security risk assessment double-layer planning model, an initial value of an upper-layer malicious attack vector is givenAnd initializing a tight constraint set;
For a given upper layer attack vectorObtaining the optimal operation state of the system corresponding to the bottom optimization problem P3Lagrange multiplier ofThe corresponding objective function value isK;
For the current upper layer attack vectorUsing the optimum operating conditions of the systemCalculating each constraint from the constraint matrixiLoad level of;
ifThen it is firstiA constraint is defined as a tight constraint storeSOtherwise, drop this constraint;is a set constant;
v. store for current iteration processSIs restricted, updatedAnd returning to step ii) until the objective function valueKIs no longer changed at this timeIs a variable ofyThe optimal value of (a), namely, the approximate optimal solution;
3) taking the optimal solution as an initial solution of a hill climbing method, and iteratively obtaining a global optimization risk value of the power network security risk assessment double-layer planning model;
wherein the bottom-level optimization problem P3 is represented as:
s.t.
2. The power network security risk assessment method according to claim 1, wherein in step 1), the expression of the power network security risk assessment double-layer planning model is as follows:
s.t.
s.t.
wherein,a risk indicator representing the power network;yrepresenting an attack vector;xrepresenting the power network operating state;A,Bis a variable ofx,yA corresponding coefficient matrix;representing a lagrange multiplier;bis a constant matrix of the lower layer constraints;Eis a variable in the upper layer constraintyA coefficient matrix of (a);gis a constant matrix of upper layer constraints.
4. The power network security risk assessment method according to claim 2, wherein in step v, the update is performed by solving P4 of the underlying model;
s.t.
Wherein,representing malicious data injected byResulting constraintiThe load increment of (a) is increased,for the current constraintiA weight factor for load increment index normalization;represents the Lagrange multiplier obtained from P3;a constant matrix representing constraints limiting the attack vector;Kan objective function value of P3; solving for P4 to obtainThen updates the next iterationThe value of (c):;representing an assignment.
5. The power network security risk assessment method according to claim 4, wherein the specific implementation process of step 3) includes:
subjecting the product of step v)Endowing a hill climbing method as an iteration initial value; for the iteration initial value, solving the bottom layer optimization problem to obtain the objective function valueKAnd lagrange multiplier;
vii. multiplying the lagrange multiplierUpdates in dual form P5 substituted into the underlying planAnd obtaining the corresponding objective function valuef;
To be updatedSubstituting into step vi), repeating steps vi) -viii) until two objective function valuesKAndfwhen none of them increases any more, orKAndfthe absolute value of the difference between the two is less than the set threshold valueIf so, considering that a system global optimization risk value is obtained, and stopping iteration; in step vii, the expression of the dual form P5 of the bottom layer plan is:
s.t.
6. the power network security risk assessment method according to claim 2,
wherein,N,NG,NDandNLthe number of nodes, the number of generators, the number of loads and the number of lines of the power system are respectively;Pfor generator output, dimension isNG×1,,ΔDAndJload data, injected false data and load shedding caused by attack, dimensions are all load data, injected false data and load shedding caused by attackND×1;PLFor line flow, its dimension isNL×1;SFIs a transfer factor matrix with dimensions ofNL×N;KP,KDAndKLrespectively represent node-generator, node-load and node-line incidence matrixes with dimensions ofN×NG,N×ND,N×NL;、Are respectively asPMinimum and maximum values of;、are respectively asPLMinimum and maximum values of.
7. A power network security risk assessment system comprises computer equipment; characterized in that the computer device is programmed or configured to perform the steps of the method for assessing the security risk of an electric power network according to any one of claims 1 to 6; or a storage medium of the computer device stores a program for executing the power network security risk assessment method according to any one of claims 1 to 6.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010392394.5A CN111641595B (en) | 2020-05-11 | 2020-05-11 | Power network security risk assessment method and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010392394.5A CN111641595B (en) | 2020-05-11 | 2020-05-11 | Power network security risk assessment method and system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111641595A CN111641595A (en) | 2020-09-08 |
CN111641595B true CN111641595B (en) | 2021-04-20 |
Family
ID=72331964
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010392394.5A Active CN111641595B (en) | 2020-05-11 | 2020-05-11 | Power network security risk assessment method and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111641595B (en) |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112260989B (en) * | 2020-09-16 | 2021-07-30 | 湖南大学 | Power system and network malicious data attack detection method, system and storage medium |
CN113283064B (en) * | 2021-05-10 | 2022-04-08 | 湖南大学 | Double-layer evaluation model, system and calculation method for network security risk of electric power system |
CN113258573B (en) * | 2021-05-10 | 2022-07-01 | 湖南大学 | Power dispatching instruction safety assessment method and system and power flow calculation method and system |
CN113516357B (en) * | 2021-05-10 | 2024-04-19 | 湖南大学 | Electric power system vulnerable line assessment method and system considering network attack risk |
CN114553517B (en) * | 2022-02-14 | 2024-05-14 | 北京源堡科技有限公司 | Nonlinear weighted network security assessment method, device, equipment and storage medium |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106845627A (en) * | 2017-01-25 | 2017-06-13 | ***股份有限公司 | A kind of data predication method and device |
CN109256225A (en) * | 2018-10-30 | 2019-01-22 | 中广核工程有限公司 | A kind of nuclear power plant containment shell lining defect detecting system, method and executive device |
CN111044808A (en) * | 2019-11-15 | 2020-04-21 | 国网江苏省电力有限公司 | Power utilization information acquisition system operation and maintenance quality reliability assessment system and method |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20160335223A1 (en) * | 2014-06-27 | 2016-11-17 | University Of South Florida | Methods and systems for computation of bilevel mixed integer programming problems |
CN104750979B (en) * | 2015-03-13 | 2017-12-15 | 上海交通大学 | The synthesization risk priority number calculating method of Architecture-oriented |
CN110097276B (en) * | 2019-04-25 | 2021-06-08 | 湖南大学 | Power grid multi-risk variable comprehensive evaluation method and system |
CN110826841A (en) * | 2019-08-31 | 2020-02-21 | 华南理工大学 | Charging station planning method considering user charging experience and power distribution network operation risk |
CN110969284B (en) * | 2019-10-29 | 2022-09-09 | 国网河南省电力公司经济技术研究院 | Double-layer optimized scheduling method for power distribution network |
-
2020
- 2020-05-11 CN CN202010392394.5A patent/CN111641595B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106845627A (en) * | 2017-01-25 | 2017-06-13 | ***股份有限公司 | A kind of data predication method and device |
CN109256225A (en) * | 2018-10-30 | 2019-01-22 | 中广核工程有限公司 | A kind of nuclear power plant containment shell lining defect detecting system, method and executive device |
CN111044808A (en) * | 2019-11-15 | 2020-04-21 | 国网江苏省电力有限公司 | Power utilization information acquisition system operation and maintenance quality reliability assessment system and method |
Also Published As
Publication number | Publication date |
---|---|
CN111641595A (en) | 2020-09-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111641595B (en) | Power network security risk assessment method and system | |
CN111262858B (en) | Network security situation prediction method based on SA _ SOA _ BP neural network | |
Tufail et al. | False data injection impact analysis in ai-based smart grid | |
Rosay et al. | Feed-forward neural network for Network Intrusion Detection | |
CN114519190A (en) | Multi-target network security dynamic evaluation method based on Bayesian network attack graph | |
CN114781692A (en) | Short-term power load prediction method and device and electronic equipment | |
Zhang et al. | Prediction algorithm for network security situation based on bp neural network optimized by sa-soa | |
CN116522747A (en) | Two-stage optimized extrusion casting process parameter optimization design method | |
CN116996272A (en) | Network security situation prediction method based on improved sparrow search algorithm | |
Sharma et al. | Anomaly based network intrusion detection for IoT attacks using convolution neural network | |
CN112651110B (en) | Malignant data injection attack defense method based on multi-stage dynamic game | |
Udas et al. | Attention-based RNN architecture for detecting multi-step cyber-attack using PSO metaheuristic | |
CN111818007B (en) | Vulnerability repair income priority evaluation method based on quantum genetic algorithm and electronic device | |
Akhavan et al. | An unsupervised feature selection for web phishing data using an evolutionary approach | |
CN117155594A (en) | Block chain self-adaptive detection method, terminal and storage medium for Sybil attack | |
CN114006744B (en) | LSTM-based power monitoring system network security situation prediction method and system | |
CN115640755A (en) | Air combat incomplete information data processing method based on extreme learning machine | |
CN115412328A (en) | Attack path tracing and attack source detection method based on machine learning | |
Yao et al. | Bayesian and stochastic game joint approach for Cross-Layer optimal defensive Decision-Making in industrial Cyber-Physical systems | |
CN114139601A (en) | Evaluation method and system for artificial intelligence algorithm model of power inspection scene | |
CN117650949B (en) | Network attack interception method and system based on RPA robot data analysis | |
Forough et al. | Unified Identification of Anomalies on the Edge: A Hybrid Sequential PGM Approach | |
CN117937521B (en) | Power system transient frequency stability prediction method, system, medium and equipment | |
CN117332923A (en) | Weighting method and system for netlike index system | |
CN117171748B (en) | Malicious code family classification method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |