A Security Control Method for Multi-Area Interconnected Power Systems under Hybrid Attacks

Abstract

The present invention relates to a security control method for a multi-region interconnected power system under hybrid attacks, comprising: establishing a cyber attack model including a false data injection attack model and a denial-of-service attack model; constructing a multi-region interconnected power system model in combination with the cyber attack model; detecting the denial-of-service attack and the false data injection attack according to the multi-region interconnected power system model, in combination with a data compensation strategy and an unscented Kalman filter algorithm, and controlling the working states of corresponding regions based on the detection results, so as to defend against the denial-of-service attack and the false data injection attack. Compared with the prior art, the present invention can accurately detect and reliably defend against both the false data injection attack and the denial-of-service attack, and effectively improve the security control effect of the multi-region interconnected power system under hybrid attacks.

Description

Technical Field

[0001] The present invention relates to the technical field of interconnected power system control, and in particular to a security control method for a multi-area interconnected power system under hybrid attacks. Background Art

[0002] With the continuous development of power grid technology, regional interconnection of power systems has been achieved, and its advantage lies in being able to integrate the resource advantages between different regions and effectively alleviate the problem of uneven distribution between load and energy. In an interconnected power system, a load change or abnormality in a certain area will cause a mismatch in the frequency of the entire power system and the interchange power of the inter-area tie lines. The key to realizing the safe and stable operation of the power grid lies in load frequency control. Generally, the network attacks on the load frequency control system are mainly false data injection attacks (FDIA) and denial of service attacks (DoS). Among them, the false data injection attack tamps the frequency and power received by the load frequency control system, causing the load frequency control system to overshoot and trigger frequency fluctuations, or even causing the frequency to cross the boundary; the denial of service attack blocks the transmission node or the transmission channel, destroying the availability of information, resulting in partial loss of information and breaking the stability of the system.

[0003] Currently, for the network attacks in the load frequency control system, there are already some different detection and control methods. Among them, the detection methods for false data injection attacks in the load frequency control system mainly include methods based on regional control error monitoring and prediction and methods based on the state observation of the load frequency control system; in the research on countermeasures against denial of service attacks, there are mainly two ideas. The first idea is to predict the lost packet data to alleviate the impact of the attack, and the second idea is to establish a robust control strategy considering denial of service attacks.

[0004] However, the above research only considers false data injection attacks or denial of service attacks. In an actual system, network attacks in the load frequency control system may occur in various control areas or transmission channels, and malicious attackers can launch multiple different attacks simultaneously; at the same time, when suffering from false data injection attacks, the power system not only needs to detect false data injection attacks, but also needs to ensure continuous safe production of electric energy without shutting down. With the increase in the degree of power grid interconnection, the scale of the power system is becoming increasingly large. The stability of the power system can be judged by the frequency of the power system, and network attacks may lead to severe oscillations in the frequency of the multi-area interconnected power system. To ensure continuous safe production of electric energy without shutting down, it is necessary to study the security control of the multi-area interconnected power system under hybrid attacks. Summary of the Invention

[0005] The object of the present invention is to overcome the defects of the above-mentioned existing technologies and provide a security control method for a multi-region interconnected power system under hybrid attacks, which can accurately detect and reliably defend against false data injection attacks and denial-of-service attacks simultaneously.

[0006] The object of the present invention can be achieved by the following technical solutions: A security control method for a multi-region interconnected power system under hybrid attacks, comprising the following steps:

[0007] S1. Establish a network attack model including a false data injection attack model and a denial-of-service attack model;

[0008] S2. Combine the network attack model established in step S1 to construct a multi-region interconnected power system model;

[0009] S3. According to the multi-region interconnected power system model, combine the data compensation strategy and the cubature Kalman filter algorithm to detect the denial-of-service attack and the false data injection attack, and control the working state of the corresponding region based on the detection results, so as to defend against the denial-of-service attack and the false data injection attack.

[0010] Further, the false data injection attack model in step S1 is specifically:

[0011]

[0012] where z k is the measurement data, x k is the state data, C is the output matrix, rank(a,b) is a random value, a is the lower bound of the random function, b is the upper bound of the random function, τ is the moment when the attacker is active, and v k is the measurement noise.

[0013] Further, the denial-of-service attack model in step S1 is specifically:

[0014]

[0015] where z k is the measurement data, x k is the state data, C is the output matrix, τ is the moment when the attacker is active, and v k is the measurement noise.

[0016] Further, the multi-region interconnected power system model in step S2 is specifically:

[0017]

[0018]

[0019]

[0020]

[0021]

[0022] Among them, Δf i (t) is the frequency deviation, ΔP gi (t) is the turbine output power, ΔX gi (t) is the governor valve position deviation, ΔP tie-i (t) is the power exchange of the tie line, ACE i (t) is the area control error, H i is the inertia coefficient, D i is the frequency sensitivity load coefficient, T Ti is the turbine time constant, R i is the speed regulation coefficient, T Gi is the time constant of the governor, B i is the frequency deviation factor, T ij is the synchronization coefficient of the tie line between area i and area j;

[0023] By simplification, the state equation of the system is obtained as:

[0024]

[0025] z(t) = Cx(t) + υ

[0026]

[0027]

[0028]

[0029] B = diag{B 1 , B 2 ,..., B N}

[0030]

[0031] C = diag{C 1 ,..., C N}

[0032] C i = [1 0 0 1 0] T

[0033] Among them, ω is the process noise, υ is the measurement noise, x(t) = [x 1 (t) x 2 (t) x 3(t)…x n (t)] T is the system state, x i (t) = [Δf i ΔP gi ΔX gi ΔP tie-i ACE i T , u = [u 1 , u 2 T is the system input, z(t) = [z1(t)z 2 (t)z 3 (t)...z n (t)] T is the system measurement, z i = [Δf i (t), ΔP tie-i (t)] T , A is the system state matrix, B is the system input matrix, and C is the system output matrix.

[0034] Further, the step S3 specifically includes the following steps:

[0035] S31. Detect a denial-of-service attack. If it is detected that the measurement data is under a denial-of-service attack, then execute step S32; otherwise, execute step S33;

[0036] S32. Adopt a data compensation strategy to compensate the measurement data, thereby defending against the denial-of-service attack, and then execute step S33;

[0037] S33. Perform state estimation on the system based on the cubature Kalman filter algorithm, and then detect whether a false data injection attack has occurred based on the estimation residual. If it is detected that a false data injection attack has occurred, then execute step S35; otherwise, execute step S34;

[0038] S34. Calculate the area control deviation using the current measurement data, and control the working state of the corresponding area accordingly, and then update the iteration count and return to step S31;

[0039] S35. Use the triple exponential smoothing method to obtain the predicted value of the area control error, update the area control deviation calculated under the false data injection attack, and control the working state of the corresponding area accordingly. Then, determine whether the preset time threshold is reached. If the determination is yes, end the current process; otherwise, update the iteration count and return to step S31.

[0040] Further, the specific process of the step S31 is:

[0041] To represent z k , z​​k-1 ,…,z k-d+1 ,z k-d The transmission status of, define the row matrix λ k , all of which follow the Bernoulli distribution property and satisfy:

[0042] Pr(λ k (i)=0)=ρ

[0043] Pr(λ k (i)=1)=1 - ρ

[0044] var(λ k (i))=ρ(1 - ρ)

[0045] Then at time k, whether z k is successfully transmitted is represented by as:

[0046]

[0047] where Pr(·) is for probability calculation, ρ is the probability value, var(·) is for variance calculation, and λ k (i) is the i-th element in λ k , used to represent the transmission status of z k-i+1 , i ∈ [1, d + 1]. When λ k (i)=0, it means that the measurement data is under a denial-of-service attack, and execute step S33; when λ k (i)=1, it means that the measurement data is not under a denial-of-service attack, and execute step S32.

[0048] Furthermore, the specific process of the said step S32 is:

[0049] To compensate for the lost data packets under a denial-of-service attack, use the latest received data packet before time k to compensate for the lost data packets under a denial-of-service attack. By defining a compensation matrix, when a denial-of-service attack occurs at time k, the compensated measurement data is:

[0050]

[0051] where M k ∈R 1×(d+1) is the compensation matrix.

[0052] Furthermore, the said step S33 specifically includes the following steps:

[0053] S331. After performing state estimation on the system based on the cubature Kalman filter algorithm, obtain the measurement residual e k at time k;

[0054] S332. Based on the measurement error ek , further calculate the error vector g k ;

[0055] S333. Compare the error vector g k with a preset threshold. If the error vector g k is greater than or equal to the preset threshold, it is determined that a false data injection attack has occurred, and step S35 is executed; otherwise, step S34 is executed.

[0056] Further, the measurement residual in step S331 is specifically:

[0057] e k = z k - z k|k-1

[0058] The error vector in step S332 is specifically:

[0059]

[0060] where cov is the covariance matrix of e k .

[0061] Further, the preset threshold in step S333 is specifically selected to be greater than the maximum norm of the residual without attack.

[0062] Compared with the prior art, the present invention has the following advantages:

[0063] First, aiming at the security control of a multi - area interconnected power system under hybrid attacks, the present invention first establishes a network attack model including false data injection attack and denial - of - service attack models, then constructs a mathematical model of the multi - area interconnected power system. Secondly, combined with a data compensation strategy and a cubature Kalman filter algorithm, it realizes the detection and defense of denial - of - service attacks and false data injection attacks. Thus, it can accurately detect and reliably defend against false data injection attacks and denial - of - service attacks at the same time, effectively improving the security control effect of the multi - area interconnected power system under hybrid attacks.

[0064] Second, the present invention realizes the defense of denial - of - service attacks by designing a data compensation strategy, effectively overcoming the adverse effects brought by denial - of - service attacks; uses the cubature Kalman filter for state estimation of the system, realizes the detection of false data injection attacks based on the estimated residual, and further updates the area control error calculated by bad data with the triple exponential smoothing predicted value, thereby realizing the defense of false data injection attacks. It not only realizes the detection of hybrid attacks, but also can perform reliable defense, ensuring that the system can continuously produce electrical energy without downtime. Description of the Drawings

[0065] Figure 1Schematic diagram of the method flow of the present invention;

[0066] Figure 2 Schematic diagram of the application process of the embodiment;

[0067] Figure 3 Load frequency control block diagram of the i-th area of the multi-area power system in the embodiment;

[0068] Figure 4 Load frequency control system block diagram with potential cyber attacks;

[0069] Figure 5 Schematic diagram of the hybrid attack detection and defense process in the embodiment;

[0070] Figure 6 Timing diagram of the denial-of-service attack in the embodiment;

[0071] Figure 7 Schematic diagram of the defense effect of the denial-of-service attack in the embodiment;

[0072] Figure 8 Δf before and after the false data injection attack in the embodiment 1 (t) comparison chart;

[0073] Figure 9 Schematic diagram of the detection of the false data injection attack in the embodiment;

[0074] Figure 10 Schematic diagram of the defense effect of the false data injection attack in the embodiment. Detailed implementation manners

[0075] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0076] Embodiment

[0077] As Figure 1 shown, a security control method for a multi-area interconnected power system under hybrid attacks includes the following steps:

[0078] S1. Establish a cyber attack model including a false data injection attack model and a denial-of-service attack model;

[0079] S2. Combine the cyber attack model established in step S1 to construct a multi-area interconnected power system model;

[0080] S3. According to the multi-area interconnected power system model, combine the data compensation strategy and the cubature Kalman filter algorithm to detect the denial-of-service attack and the false data injection attack, and control the working states of the corresponding areas based on the detection results, so as to defend against the denial-of-service attack and the false data injection attack.

[0081] This embodiment applies the above technical solution. As Figure 2 shown, the main contents are as follows:

[0082] I. Construct a load frequency control model for a multi - area interconnected power system

[0083] In this embodiment, the load frequency control block diagram of the i - th area of the multi - area interconnected power system is as Figure 3 shown. The mathematical model of this system is:

[0084]

[0085]

[0086]

[0087]

[0088]

[0089] In the formula, Δf i (t) is the frequency deviation, ΔP gi (t) is the turbine output power, ΔX gi (t) is the governor valve position deviation, ΔP tie -i(t) is the power exchange of the tie line, ACE i (t) is the area control error, H i is the inertia coefficient, D i is the frequency sensitivity load coefficient, T Ti is the turbine time constant, R i is the governor coefficient, T Gi is the governor time constant, B i is the frequency deviation factor, T ij is the synchronization coefficient of the tie line between area i and area j. Simplifying the above formula, the state equation of the system is obtained as:

[0090]

[0091] z(t) = Cx(t)+υ

[0092] where ω is the process noise, υ is the measurement noise, x(t)=[x 1 (t) x 2 (t) x 3 (t) … x n (t)] T is the system state, where x i (t)=[Δf i ΔP gi ΔXgi ΔP tie-i ACE i ] T ,u=[u1,u 2 ] T is the system input, z(t)=[z 1 (t)z 2 (t)z 3 (t)…z n (t)] T is the system measurement, where z i =[Δf i (t),ΔP tie-i (t)] T , the system parameters are as follows:

[0093]

[0094]

[0095]

[0096] B = diag{B 1 ,B 2 ,…,B N}

[0097]

[0098] C=diag{C 1 ,...,C N}

[0099] C i =[1 0 0 1 0] T

[0100] Among them, A is the system state matrix, B is the system input matrix, and C is the system output matrix.

[0101] II. Network Attack Model

[0102] The system block diagram of the load frequency control system that is attacked by the network is as follows: Figure 4 As shown by Figure 4 It can be seen that the attacker will carry out false data injection attacks on the frequency sensor and the tie line power sensor, and the attacker will also carry out denial of service attacks on the transmission channels of frequency measurement and tie line power measurement.

[0103] 2.1 False Data Injection Attack Model

[0104] Consider the false data injection attack model as a random attack, which adds a random value rank (a, b) to the measurement z k In, making

[0105]

[0106] where a is the lower bound of the random function, b is the upper bound of the random function, τ is the moment when the attacker is active, and v k is the measurement noise.

[0107] 2.2 Denial-of-Service Attack Model

[0108] To interfere with the system's communication within a certain period of time, the denial-of-service attack will block the transmission channel or the transmission node, resulting in the loss of the measurement data z k packets, satisfying the following formula:

[0109]

[0110] III. Detection and Defense of Network Attacks

[0111] The control process of the multi-area interconnected power system based on cubature Kalman filter under hybrid attacks is as Figure 5 shown.

[0112] 3.1 Detection of Denial-of-Service Attacks

[0113] To represent the transmission status of z k , z k-1 , …, z k-d+1 , z k-d define the row matrix λ k , which all follow the Bernoulli distribution property and satisfy:

[0114] Pr(λ k (i) = 0) = ρ

[0115] Pr(λ k (i) = 1) = 1 - ρ

[0116] var(λ k (i)) = ρ(1 - ρ)

[0117] Pr(·) represents probability calculation, var(·) represents variance calculation, and λ k (i) is the i-th element of λ k , which represents the transmission status of z k-i+1 , i ∈ [1, d + 1]. When λ k (i) = 0, it means that the measurement data is under a DoS attack. When λ k (i) = 1, it means that the measurement data is not under a DoS attack.

[0118] Whether the transmission of z k is successful at time k is represented by as

[0119]

[0120] 3.2 Defense Against Denial-of-Service Attacks

[0121] To compensate for the lost data packets under a denial-of-service attack, the most recently received data packet before time k is used to compensate for the lost data packets under the denial-of-service attack. Define the compensation matrix M k ∈R 1×(d+1) When a denial-of-service attack occurs at time k, the compensated measurement data is

[0122]

[0123] 3.3 Detection of False Data Injection Attacks

[0124] The unscented Kalman filter algorithm is used to perform state estimation on the system, and then the detection of false data injection attacks is realized based on the estimation residuals. After performing state estimation on the system based on the unscented Kalman filter algorithm, the measurement residual e at time k k is

[0125] e k = z k - z k|k-1

[0126] To detect false data injection attacks, define the error vector g k as

[0127]

[0128] where cov is the covariance matrix of e k .

[0129] The detector compares g k with a predefined threshold, which is chosen to be greater than the maximum norm of the residuals in the absence of attacks.

[0130] 3.4 Defense Against False Data Injection Attacks

[0131] The triple exponential smoothing method is used to predict the area control error. When a false data injection attack is detected in the measurement data, the triple exponential smoothing predicted value is used to update the area control error calculated from the bad data, which can weaken the impact of false data injection attacks on the power balance and frequency stability of the power system.

[0132] IV. Aiming at detecting and defending against hybrid attacks, determine the security control strategy of the multi-area interconnected power system.

[0133] In this embodiment, assume that the probability of measurement data loss is 0.06, and the corresponding data packet loss time sequence diagram is as shown in Figure 6as shown, where "0" indicates packet loss and "1" indicates normal packet transmission. From Figure 7 It can be seen that when a denial-of-service attack occurs, the proposed denial-of-service attack defense method can effectively suppress the impact of the DoS attack on the tie-line power.

[0134] When a random false data injection attack is injected into the frequency deviation in Area 1, from Figure 8 it can be seen that the added false data injection attack will affect the frequency stability in Area 1. From Figure 9 it can be seen that the proposed detection method can effectively detect the added false data injection attack. When a false data injection attack is detected, the area control error calculated from the bad data is updated by the area control error predicted by the triple exponential smoothing method. From Figure 10 the effectiveness of the proposed false data injection attack defense method is verified.

[0135] In summary, for the security control problem of a multi-area interconnected power system, the proposed technical solution fully considers the detection and defense of hybrid attacks, and can accurately detect and reliably defend against false data injection attacks and denial-of-service attacks.

Claims

1. A security control method for a multi - area interconnected power system under hybrid attacks, characterized in that, it includes the following steps: S1. Establish a network attack model including a false data injection attack model and a denial - of - service attack model; S2. Combine the network attack model established in step S1 to construct a multi - area interconnected power system model; S3. According to the multi - area interconnected power system model, combine the data compensation strategy and the cubature Kalman filter algorithm to detect the denial - of - service attack and the false data injection attack, and control the working state of the corresponding area based on the detection results, so as to defend against the denial - of - service attack and the false data injection attack; Step S3 specifically includes the following steps: S31. Detect the denial - of - service attack. If it is detected that the measurement data is under a denial - of - service attack, then execute step S32; otherwise, execute step S33; S32. Adopt the data compensation strategy to compensate the measurement data, so as to defend against the denial - of - service attack, and then execute step S33; S33. Perform state estimation on the system based on the cubature Kalman filter algorithm, and then detect whether a false data injection attack has occurred based on the estimated residual. If it is detected that a false data injection attack has occurred, then execute step S35; otherwise, execute step S34; S34. Calculate the area control deviation using the current measurement data, and control the working state of the corresponding area based on this, then update the iteration count, and return to step S31; S35. Adopt the triple exponential smoothing method to obtain the predicted value of the area control error, update the area control deviation calculated under the false data injection attack, and control the working state of the corresponding area based on this. Then, determine whether the preset time threshold has been reached. If it is judged to be yes, end the current process; otherwise, update the iteration count and return to step S31.

2. According to the security control method for a multi - area interconnected power system under hybrid attacks described in claim 1, characterized in that, the false data injection attack model in step S1 is specifically: where z k is measurement data, x k is status data, C is the output matrix, rank(a, b) is a random value, a is the lower bound of the random function, b is the upper bound of the random function, τ is the instant when the attacker is active, v k is measurement noise.

3. According to the security control method for a multi - area interconnected power system under hybrid attacks described in claim 2, characterized in that, the denial - of - service attack model in step S1 is specifically: where z k is the measurement data, x k is the state data, C is the output matrix, τ is the instant when the attacker is active, and v k is the measurement noise.

4. According to the security control method for a multi - area interconnected power system under hybrid attacks described in claim 3, characterized in that, the multi - area interconnected power system model in step S2 is specifically: Among them, Δf i (t) is the frequency deviation, ΔP gi (t) is the turbine output power, ΔX gi (t) is the governor valve position deviation, ΔP tie-i (t) is the power exchange of the tie line, ACE i (t) is the area control error, H i is the inertia coefficient, D i is the frequency sensitivity load coefficient, T Ti is the turbine time constant, R i is the speed regulation coefficient, T Gi is the time constant of the governor, B i is the frequency deviation factor, T ij is the synchronizing coefficient of the tie line between area i and area j; By simplification, the state equation of the system is obtained as: z(t) = Cx(t)+υ B = diag{B 1 , B 2 , …, B N} C = diag{C 1 , …, C N} C i ＝[10010] T where ω is the process noise, υ is the measurement noise, and x(t) = [x 1 (t)x 2 (t)x 3 (t)L x n (t)] T is the system state, x i (t) = [Δf i ΔP gi ΔX gi ΔP tie-i ACE i T , u = [u 1 , u 2 T is the system input, z(t) = [z 1 (t)z 2 (t)z 3 (t)Lz n (t)] T is the system measurement, z i = [Δf i (t), ΔP tie-i (t)] T , A is the system state matrix, B is the system input matrix, and C is the system output matrix.​​ 5. According to the security control method for a multi - area interconnected power system under hybrid attacks described in claim 1, characterized in that, the specific process of step S31 is: To represent z k , z k-1 , …, z k-d+1 , z k-d For the transmission status of, define the row matrix λ k , all of which follow the Bernoulli distribution property and satisfy: Pr(λ k (i)=0)=ρ Pr(λ k (i)=1)=1 - ρ var(λ k (i)) = ρ(1 - ρ) Then at time k, z k Whether the transmission is successful is indicated by expressed as: Among them, Pr(·) is the probability calculation, ρ is the probability value, var(·) is the variance calculation, and λ k (i) is the i-th element of λ k and is used to represent the transmission state of z k-i+1 , where i ∈ [1, d + 1]. When λ k (i) = 0, it means that the measurement data is under a denial-of-service attack, and step S33 is executed; when λ k (i) = 1, it means that the measurement data is not under a denial-of-service attack, and step S32 is executed.

6. According to the security control method for a multi - area interconnected power system under hybrid attacks described in claim 5, characterized in that, the specific process of step S32 is: To compensate for the data packets lost under the denial - of - service attack, use the latest received data packet before the k - th moment to compensate for the data packets lost under the denial - of - service attack. By defining the compensation matrix, when a denial - of - service attack occurs at the k - th moment, the compensated measurement data is: where M k ∈R 1×(d+1) is the compensation matrix.

7. According to the security control method for a multi - area interconnected power system under hybrid attacks described in claim 1, characterized in that, The specific steps of step S33 include the following steps: After performing state estimation on the system based on the volume Kalman filter algorithm, the measurement residual e at time k is obtained k ; S332. Based on the measurement error e k , further calculate the error vector g k ; S333. Compare the error vector g k with a preset threshold. If the error vector g k is greater than or equal to the preset threshold, it is determined that a false data injection attack has occurred and step S35 is executed; otherwise, step S34 is executed.

8. A security control method for a multi-region interconnected power system under hybrid attacks according to claim 7, characterized in that, the measurement residual in step S331 is specifically: e k = z k -z k|k-1 the error vector in step S332 is specifically: Among them, cov is the covariance matrix of e k .

9. A security control method for a multi-region interconnected power system under hybrid attacks according to claim 7, characterized in that, the preset threshold in step S333 is specifically selected to be greater than the maximum norm of the residual without attacks.

Images