Attack detection and elastic control method for multi-region power market coupling system

Abstract

The invention relates to the technical field of power system network security and load frequency control, in particular to an attack detection and elastic control method for a multi-region power market coupling system, which comprises the following steps of: establishing a multi-region power market coupling system model containing a multi-channel FDIA; the method comprises the following steps: respectively modeling FDIAs of a control channel and a measurement channel as unknown input and unknown output of a system, constructing an augmented system model containing an original system state and an attack signal, designing a state and attack observer, synchronously estimating the system state and the attack signal by using local measurement data, judging whether each regional channel is attacked or not, and if yes, judging whether each regional channel is attacked or not. When it is judged that the load frequency control LFC controller is attacked, the load frequency control LFC controller is compensated, and the load frequency control LFC controller is reconstructed; according to the method, effective defense of multi-region multi-channel FDIA is realized through synchronous attack estimation and compensation.

Description

Technical Field

[0001] This invention relates to the field of power system network security and load frequency control technology, and in particular to an attack detection and resilient control method for a multi-regional power market coupled system. Background Technology

[0002] With the continuous improvement of the intelligence level of power systems, cyber-physical integration has become a typical feature of modern power grids. Automatic generation control (AGC), as a key system for maintaining grid frequency stability, highly depends on the reliability of the communication network. However, security flaws in standard SCADA communication protocols, such as those related to authentication and patch management, pose serious cybersecurity threats to power systems.

[0003] False Data Injection (FDIA) attacks can disrupt the normal operation of a system and even trigger cascading failures by tampering with measurement data or control commands. Traditional attack defense methods have significant limitations: data-driven methods require large amounts of training data and have poor robustness to changes in model parameters; model-driven methods mostly only detect attacks but cannot effectively compensate for them, and their centralized architecture makes them difficult to handle multi-region, multi-channel attacks; methods based on hardware redundancy increase system cost and complexity. To address these limitations, this invention aims to develop a multi-region power system FDIA detection and resilient control method based on State and Attack Observer (SAO), achieving effective defense against multi-region, multi-channel FDIA through synchronous attack estimation and compensation. Summary of the Invention

[0004] This invention provides an attack detection and resilient control method for a multi-regional power market coupling system, which achieves effective defense against multi-regional and multi-channel FDIA through synchronous attack estimation and compensation.

[0005] To achieve the above objectives, the present invention employs the following technical solution: A method for attack detection and resilient control of a multi-regional electricity market coupling system includes the following steps: S1. Establish a coupled system model of a multi-regional electricity market with multi-channel FDIA, and model the FDIA of the control channel and the measurement channel as unknown input and unknown output of the system, respectively. The coupled system model is as follows: ; in, For the system matrix, For the input matrix, For the region coupling matrix, To control the channel attack matrix, For the output matrix, To measure the channel attack matrix, This is the inter-region coupling matrix. Let be the time derivative of the state vector of the i-th region. Let be the state vector of the i-th region. This is the control input for the i-th region. Inject attack signals into the false data of the control channel in the i-th region. Injecting attack signals into the false data of the measurement channel in the i-th region. Let be the robustness performance weighting coefficient for the i-th region. This is the control output for the i-th region; S2. Construct an augmented system model that includes the original system state and attack signals: ; in, Let be the state vector of the augmented system. To augment the time derivative of the system's state vector, , , It is the identity matrix. ; S3. Design a status and attack observer to synchronously estimate the system status and attack signals using local measurement data. It determines whether each area channel is under attack. If the estimated value of the attack signal is 0, then the system is not under attack; otherwise, it is under attack. S4. When it is determined that an attack has been suffered, the load frequency control LFC controller is compensated and reconstructed.

[0006] Furthermore, the establishment of a multi-regional electricity market coupling system model including multi-channel FDIA specifically includes the following process: Based on producer and consumer behavior and market dynamics, the dynamic mechanism of the electricity market is represented as follows: ; in, Power supply for producers For the power demand of consumers, For electricity prices, and These are the producer's marginal cost and the consumer's marginal revenue, respectively. and These are the producer's response constant and the consumer's response constant, respectively. Let be the electricity price response rate constant. For the benefit of market stabilizers of energy imbalances, The system is in an energy imbalance. The marginal cost intercept for producers. The slope of the producer's marginal cost. For the consumer's marginal revenue intercept, Let be the slope of marginal revenue for consumers. The rate of change of power supply to producers, The rate of change in consumer power demand. The rate of change in electricity prices; The system dynamics of the i-th region are represented as follows: ; in, For frequency deviation, For mechanical power deviation, To adjust the valve position deviation, For tie line power deviation, , , , , , These are, respectively, the inertial constant, damping coefficient, generator time constant, governor time constant, speed regulation coefficient, and synchronization coefficient between the i-th and j-th regions. For load disturbance, The output of the LFC controller is used to control the load frequency. Load Frequency Control LFC Controller Output for: ; in, It is the frequency deviation factor. and These are proportional gain and integral gain. For regional control deviation; Define the state vector of the i-th region as The control input for the i-th region is The control output of the i-th region ; The multi-regional electricity market coupling system model considering multi-channel FDIA and model uncertainties is as follows: ; in, .

[0007] Furthermore, the design state and attack observer specifically include the following processes: The state and attack observers are represented as follows: ; in, Indicates the state of the estimator. Original system state The estimated value, , These are attack signals. and The estimated value, By choosing the gain matrix, the matrix becomes It is neither strange nor unusual. To augment the nonsingular gain matrix of the system, To compensate for the output derivative gain matrix, The observer gain matrix is... To control the time derivative of the output, The time derivative of the estimator state; To eliminate Introducing new observer states Then the state and attack observer are: ; in, For the new observer state, The time derivative of the new observer state; because , The final state and attack observer are obtained: .

[0008] Furthermore, step S4 specifically includes the following process: Using estimated values Compensation due to attack signals on the measurement channel Affected The compensated measurement signal: ; in, Attack signals on the measurement channel The estimation error; Area control error: ; in, ; When it is determined that the i-th region has been attacked, the control quantity is adjusted. Compensation is performed, that is, compensation is applied to the load frequency control LFC controller, and the load frequency control LFC controller is reconstructed: ; definition , These are the state measurement error and the control channel attack signal measurement error, respectively. The estimation error is augmented to... Dynamically represented as , ;; in, The output of the compensated LFC controller, To augment the estimation error, The time derivative of the augmented estimation error.

[0009] Compared with the prior art, the beneficial effects of the present invention are: Effective defense against multi-region, multi-channel FDIA is achieved through synchronous attack estimation and compensation. Each region's SAO and controller rely only on local data, without requiring global system information, thus avoiding the single-point failure risk of centralized architecture and handling multi-region attacks. Simultaneously, FDIA of control and measurement channels is detected and compensated, overcoming the shortcomings of existing methods that only defend against single-channel attacks. The impact of model uncertainty is suppressed, enabling adaptation to changes in the operating conditions of the power market coupled system. Attached Figure Description

[0010] Figure 1 This refers to the frequency deviation of region 1 under a single-region, single-channel attack scenario in this embodiment of the invention. Estimated simulation diagram.

[0011] Figure 2 This refers to the power deviation of the region 1 tie-line under a single-region, single-channel attack scenario in this embodiment of the invention. Estimated simulation diagram.

[0012] Figure 3 This is an attack signal against region 1 under the single-region, single-channel attack scenario in this embodiment of the invention. Estimated simulation diagram.

[0013] Figure 4 This refers to region 1 in the single-region, single-channel attack scenario of this embodiment of the invention. Response simulation diagram.

[0014] Figure 5 This refers to the region 2 frequency deviation under a single-region, single-channel attack scenario in this embodiment of the invention. Simulation diagram.

[0015] Figure 6 This refers to the region 3 frequency deviation under a single-region, single-channel attack scenario in this embodiment of the invention. Simulation diagram.

[0016] Figure 7 This is the region 1 attack signal under the single-region multi-channel attack scenario in this embodiment of the invention. Simulation comparison of SAO estimation and KF estimation.

[0017] Figure 8 This is the region 1 attack signal under the single-region multi-channel attack scenario in this embodiment of the invention. SAO estimation simulation diagram.

[0018] Figure 9 This refers to region 1 in the single-region multi-channel attack scenario of this invention embodiment. Simulation comparison of responses under SAO elastic control and KF control.

[0019] Figure 10 This refers to the region 2 frequency deviation under a single-region multi-channel attack scenario in this embodiment of the invention. Simulation comparison of responses under SAO elastic control and KF control.

[0020] Figure 11 This refers to the region 3 frequency deviation under a single-region multi-channel attack scenario in this embodiment of the invention. Simulation diagrams under SAO elastic control and KF control.

[0021] Figure 12 This is the region 1 attack signal under the multi-region, multi-channel attack scenario in this embodiment of the invention. Simulation comparison of SAO estimation and KF estimation.

[0022] Figure 13 This is the region 2 attack signal under the multi-region, multi-channel attack scenario in this embodiment of the invention. Simulation comparison of SAO estimation and KF estimation.

[0023] Figure 14 This is the region 2 attack signal under the multi-region, multi-channel attack scenario in this embodiment of the invention. SAO estimation simulation diagram.

[0024] Figure 15 This refers to region 1 in the multi-region, multi-channel attack scenario of this invention embodiment. Simulation comparison of responses under SAO elastic control and KF control.

[0025] Figure 16 This refers to the region 2 frequency deviation under multi-region, multi-channel attack conditions in this embodiment of the invention. Simulation comparison of responses under SAO elastic control and KF control.

[0026] Figure 17This refers to the region 3 frequency deviation under multi-region, multi-channel attack conditions in this embodiment of the invention. Simulation comparison of responses under SAO elastic control and KF control. Detailed Implementation

[0027] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings: Explanation of mathematical symbols: This represents the transpose operation of a matrix. The operation represents the inversion of a matrix. This represents the summation operation. It is a collection of multi-regional power systems.

[0028] The following embodiments are implemented based on the technical solution of the present invention, providing detailed implementation methods and specific operation processes. However, the scope of protection of the present invention is not limited to the following embodiments. Unless otherwise specified, the methods used in the following embodiments are conventional methods.

[0029] Example 1 This invention discloses an attack detection and resilient control method for a multi-regional electricity market coupling system, comprising the following steps: S1. Establish a multi-regional power market coupled system model with multi-channel FDIA, and model the FDIA of the control channel and the measurement channel as unknown input and unknown output of the system, respectively; Based on producer and consumer behavior and market dynamics, the dynamic mechanism of the electricity market can be represented by the following equation: (1) in, Power supply for producers For the power demand of consumers, For electricity prices, and These are the producer's marginal cost and the consumer's marginal revenue, respectively. and These are the producer's response constant and the consumer's response constant, respectively. Let be the electricity price response rate constant. For the benefit of market stabilizers of energy imbalances, The system is in an energy imbalance. The marginal cost intercept for producers. The slope of the producer's marginal cost. For the consumer's marginal revenue intercept, Let be the slope of marginal revenue for consumers. The rate of change of power supply to producers, The rate of change in consumer power demand. The rate of change in electricity prices; The system dynamics of the i-th region can be represented in the following form: (2) in, For frequency deviation, For mechanical power deviation, To adjust the valve position deviation, For tie line power deviation, , , , , , These are, respectively, the inertial constant, damping coefficient, generator time constant, governor time constant, speed regulation coefficient, and synchronization coefficient between the i-th and j-th regions. For load disturbance, The output of the LFC controller is used to control the load frequency. Based on traditional PI control, the LFC controller output is controlled by load frequency. for: (3) in, It is the frequency deviation factor. and These are proportional gain and integral gain. For regional control deviation; Define the state vector of the i-th region as The control input for the i-th region is The control output of the i-th region ,in For LFC controller output, For load disturbances, the spoofed data injection attack in the i-th region is divided into attacks on the control channel. and attacks on measurement channels ; The multi-regional electricity market coupling system model considering multi-channel FDIA and model uncertainties is as follows: (4) in, For the system matrix, For the input matrix, For the region coupling matrix, To control the channel attack matrix, For the output matrix, To measure the channel attack matrix, This is the inter-region coupling matrix. Let be the time derivative of the state vector of the i-th region. Let be the state vector of the i-th region. This is the control input for the i-th region. Inject attack signals into the false data of the control channel in the i-th region. Injecting attack signals into the false data of the measurement channel in the i-th region. Let be the robustness performance weighting coefficient for the i-th region. This is the control output for the i-th region; Specifically: .

[0030] S2. Construct an augmented system model that includes the original system state and attack signals: To estimate the attack on the system, a new system state vector is defined. The original system is extended into an augmented system model that includes attack signals: (5) in, Let be the state vector of the augmented system. To augment the time derivative of the system's state vector, , , It is the identity matrix. ; S3, Design Status and Attack Observer, uses local measurement data to synchronously estimate system status and attack signals, and determines whether each area channel is under attack; (6) in, Indicates the state of the estimator. Original system state The estimated value, , These are attack signals. and The estimated value, By choosing the gain matrix, the matrix becomes It is neither strange nor unusual. To augment the nonsingular gain matrix of the system, To compensate for the output derivative gain matrix, The observer gain matrix is... To control the time derivative of the output, The time derivative of the estimator state; To eliminate Introducing new observer states Then the state and attack observer are: (7) in, For the new observer state, The time derivative of the new observer state; because , The final state and attack observer are obtained: (8) Based on the attack estimates of the control channel and measurement channel, determine whether each area channel has been attacked. If the estimated value of the attack signal is 0, then it has not been attacked; otherwise, it has been attacked.

[0031] S4. When it is determined that an attack has been suffered, the load frequency control LFC controller is compensated and reconstructed. Using estimated values Compensation due to attack signals on the measurement channel Affected The compensated measurement signal: (9) in, Attack signals on the measurement channel The estimation error; from equations (3) and (9), the regional control error can be obtained: (10) in, ; Next, following the method described above, when it is determined that the i-th region has been attacked, the control quantity is adjusted. Provide compensation: (11) definition , These are the state measurement error and the control channel attack signal measurement error, respectively. The estimation error is augmented to... Its dynamic representation is , ; in, The output of the compensated LFC controller, To augment the estimation error, The time derivative of the augmented estimation error.

[0032] The multi-regional power market coupling system is designed with multiple parallel-controlled SAOs. Each SAO only needs to perform local measurements for each region and does not involve state measurements of other regions.

[0033] For the i-th power system control area, in the presence of model uncertainty, the SAO is designed to simultaneously estimate the attack signal according to equation (8). and The attack estimates provided by SAO reveal attacks on each communication channel, thereby enabling FDIA detection for each control area.

[0034] According to equation (9), based on the estimated signal Compensated output signal According to equation (11), based on the estimated signal Compensation control signal .

[0035] Example 2 The specific implementation steps are the same as in Example 1, but a three-region power system is used as the simulation object. The parameters of the three-region power system are shown in Table 1. Table 1 area <imgwi="5.25"he="6.60"file="GrG3HpPzflET60Tk2mvZLZFentdCFhYqPul5ZCdS.jpg"img-format="jpg"img-content="drawing"orientation="portrait"inline="no"> <imgwi="5.25"he="6.60"file="pSDMtnuf7f0bx9IGL3QVQ6mBuP2ntWiTLI3GPDk8.jpg"img-format="jpg"img-content="drawing"orientation="portrait"inline="no"> <imgwi="5.25"he="6.60"file="nEo69B4lhkYhwx5qWtp5nhXz5X2fJN5M0gEcPjIU.jpg"img-format="jpg"img-content="drawing"orientation="portrait"inline="no"> <imgwi="4.74"he="6.60"file="n2p0vxDjrbmn6TA969YDfY0KIQgM77NtKDJIhpJ0.jpg"img-format="jpg"img-content="drawing"orientation="portrait"inline="no"> <imgwi="5.25"he="6.60"file="7Gikfao8UKwBGR3wJ919RyB8VPrbMRjT64jB8pSc.jpg"img-format="jpg"img-content="drawing"orientation="portrait"inline="no"> <imgwi="5.25"he="6.60"file="qA8t4IfvkBYbf0DMaA5A3jVBOf5tsoBUrmBRzgFW.jpg"img-format="jpg"img-content="drawing"orientation="portrait"inline="no"> 1 10 1 0.3 0.05 0.1 0.8 35 -0.8 150 60 2 12 1.5 0.4 0.05 0.17 0.7 30 -0.7 150 9 3 12 1.8 0.35 0.05 0.20 0.7 25 -0.6 150 48 Select interconnection parameters , , , .

[0036] The simulation starts from steady state, and... Time zone 1 appears The performance of the patent was evaluated under load perturbation conditions, both single and multiple attack conditions, and compared with KF-based solutions under the same conditions. Scenario 1: Consider the case of a single region and single channel being attacked by FDI; the injection amplitude in the measurement channel of region 1 is... The pulse attack, the attack time is The pulse attack on the measurement channel tampered with the AGC's feedback information, leading to... The steady-state error; furthermore, the output of region 1 further affects the operation of regions 2 and 3, leading to , The decline; See Figure 1-6 The observer designed using this method can accurately estimate the system state. , and attack signals Apply a flexible control scheme based on the estimated value. By compensating for AGC and reconfiguring the controller, the system output reached steady state more quickly. , , The response is close to the system state under normal conditions.

[0037] Scenario 2: Consider the case where multiple channels in a single area are attacked by FDI; At that time, the amplitude injected into the measurement channel of region 1 was... A step attack; At that time, inject into the control channel of Zone 1 Scaling attacks; See Figure 7-11 ,exist During this period, both the Kalman filter-based observer and the SAO were able to track... The true values ​​of all parameters can detect the FDIA in region 1 and compensate for the attack; after 25 seconds, the estimated values ​​of the observer based on Kalman filtering... The large estimation error compared to the true value indicates that this method cannot cope with multi-channel attacks; while SAO can simultaneously estimate the attack signals of the measurement channel and the control channel. The elastic control scheme proposed in this patent compensates for multi-channel attacks. , , The response quickly returned to normal.

[0038] Scenario 3: Consider the situation where multiple regions and multiple channels are attacked by FDI. At that time, the amplitude injected into the measurement channel of region 1 was... A step attack; At that time, the slope injected into the measurement channel of region 2 was... The slope attack; At that time, an amplitude of [value] is injected into the control channel of region 2. A sinusoidal attack with a frequency of 1.

[0039] See Figure 12-17 Kalman filter-based observers cannot track and The true value of the Kalman filter is affected by the mutual influence between the KF estimation results of region 1 and region 2, leading to a large estimation error; based on the Kalman filter algorithm... , , The response exhibits significant oscillations and cannot solve the detection and compensation problem of multi-region FDIA. Applying this patented solution, SAO can simultaneously estimate... , and Utilizing a flexible control scheme to compensate for multi-channel attacks. , , The response was close to normal, and the power system in the three regions entered a steady state, proving the feasibility of the plan.

Claims

1. A method for attack detection and resilient control of a multi-regional electricity market coupling system, characterized in that, Includes the following steps: S1. Establish a coupled system model of a multi-regional electricity market with multi-channel FDIA, and model the FDIA of the control channel and the measurement channel as unknown input and unknown output of the system, respectively. The coupled system model is as follows: ； in, For the system matrix, For the input matrix, For the region coupling matrix, To control the channel attack matrix, For the output matrix, To measure the channel attack matrix, This is the inter-region coupling matrix. Let be the time derivative of the state vector of the i-th region. Let be the state vector of the i-th region. This is the control input for the i-th region. Inject attack signals into the false data of the control channel in the i-th region. Injecting attack signals into the false data of the measurement channel in the i-th region. Let be the robustness performance weighting coefficient for the i-th region. This is the control output for the i-th region; S2. Construct an augmented system model that includes the original system state and attack signals: ； in, Let be the state vector of the augmented system. To augment the time derivative of the system's state vector, , , It is the identity matrix. ； S3. Design a status and attack observer to synchronously estimate the system status and attack signals using local measurement data. It determines whether each area channel is under attack. If the estimated value of the attack signal is 0, then the system is not under attack; otherwise, it is under attack. S4. When it is determined that an attack has been suffered, the load frequency control LFC controller is compensated and reconstructed.

2. The attack detection and resilient control method for a multi-regional electricity market coupling system according to claim 1, characterized in that, The establishment of a multi-regional electricity market coupling system model including multi-channel FDIA specifically includes the following process: Based on producer and consumer behavior and market dynamics, the dynamic mechanism of the electricity market is represented as follows: ； in, Power supply for producers For the power demand of consumers, For electricity prices, and These are the producer's marginal cost and the consumer's marginal revenue, respectively. and These are the producer's response constant and the consumer's response constant, respectively. Let be the electricity price response rate constant. For the benefit of market stabilizers of energy imbalances, The system is in an energy imbalance. The marginal cost intercept for producers. The slope of the producer's marginal cost. For the consumer's marginal revenue intercept, Let be the slope of marginal revenue for consumers. The rate of change of power supply to producers, The rate of change in consumer power demand. The rate of change in electricity prices; The system dynamics of the i-th region are represented as follows: ； in, For frequency deviation, For mechanical power deviation, To adjust the valve position deviation, For tie line power deviation, , , , , , These are, respectively, the inertial constant, damping coefficient, generator time constant, governor time constant, speed regulation coefficient, and synchronization coefficient between the i-th and j-th regions. For load disturbance, The output of the LFC controller is used to control the load frequency. Load Frequency Control LFC Controller Output for: ； in, It is the frequency deviation factor. and These are proportional gain and integral gain. For regional control deviation; Define the state vector of the i-th region as The control input for the i-th region is The control output of the i-th region ; The multi-regional electricity market coupling system model considering multi-channel FDIA and model uncertainties is as follows: ； in, 。 3. The attack detection and resilient control method for a multi-regional electricity market coupling system according to claim 1, characterized in that, The design state and attack observer specifically include the following processes: The state and attack observers are represented as follows: ； in, Indicates the state of the estimator. Original system state The estimated value, , These are attack signals. and The estimated value, By choosing the gain matrix, the matrix becomes It is neither strange nor unusual. To augment the nonsingular gain matrix of the system, To compensate for the output derivative gain matrix, The observer gain matrix is... To control the time derivative of the output, The time derivative of the estimator state; To eliminate Introducing new observer states Then the state and attack observer are: ； in, For the new observer state, The time derivative of the new observer state; because , The final state and attack observer are obtained: 。 4. The attack detection and resilient control method for a multi-regional electricity market coupling system according to claim 1, characterized in that, Step S4 specifically includes the following process: Using estimated values Compensation due to attack signals on the measurement channel Affected The compensated measurement signal: ； in, Attack signals on the measurement channel The estimation error; Area control error: ; in, ; When it is determined that the i-th region has been attacked, the control quantity is adjusted. Compensation is performed, that is, compensation is applied to the load frequency control LFC controller, and the load frequency control LFC controller is reconstructed: ； definition , These are the state measurement error and the control channel attack signal measurement error, respectively. The estimation error is augmented to... Dynamically represented as , ;; in, The output of the compensated LFC controller, To augment the estimation error, The time derivative of the augmented estimation error.

Images