End-side cloud collaborative industrial control system state variable regulation and control method

Through the state variable control method of end-edge cloud collaboration, the problem that the industrial control system cannot detect external data tampering is solved, and the effect of deviating from the ideal state trajectory to the greatest extent under finite energy is achieved, and the safety protection capability of the industrial control system is improved.

CN119087861AActive Publication Date: 2024-12-06HARBIN INST OF TECH
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202411209232.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-30
Publication Date
2024-12-06
Estimated Expiration
2044-08-30

AI Technical Summary

Technical Problem

The existing industrial control system cannot effectively detect and prevent external data tampering, causing the system status to deviate from the expected trajectory, causing security risks and potential accidents.

Method used

Using the state variable regulation method of end-edge cloud collaboration, by building a nominal system and residual generator on a cloud server, calculating the concealment matrix and state optimal estimates, generating a state offset or regulating data sequence, and implementing it on edge nodes Data injection realizes the status control of the controlled system.

Benefits of technology

Under the constraints of finite energy, it will deviate from the ideal state trajectory to the greatest extent, provide strong anti-training opponents, conceal the system status, avoid the alarm of the original abnormality detection mechanism, and improve the safety protection capabilities of the industrial control system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119087861A_ABST
    Figure CN119087861A_ABST
Patent Text Reader

Abstract

The invention discloses an end-side cloud collaborative industrial control system state variable regulation and control method, and belongs to the technical field of industrial control system automatic monitoring and industrial process safety. The invention aims to solve the problems that the change of a data track cannot be effectively found due to the fact that an existing industrial control system based on data driving does not have a state variable regulation and control method, and the defects of an automatic monitoring technology and an industrial process fault diagnosis technology exist. According to the method, online data is collected through a terminal collection and injection system, a state regulation and control data sequence is injected, data processing and small-scale real-time calculation are carried out when the data passes through an edge node, historical data and a system model are stored through a cloud server, and a complex formula and a large matrix are calculated off line; through cooperation of end-side clouds of the three, an attack sequence enabling the state track offset to be maximum is generated in the online deployment process, or a state manipulation attack sequence is generated; and performing data injection according to different sequences.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of industrial control system automation monitoring and industrial process safety, and specifically relates to a method for regulating and controlling state variables of an industrial control system. Background Art

[0002] The long-term safe and stable operation of industrial processes benefits from the development of industrial control security technology and the application of automated monitoring methods. In the new industrial context, a large number of intelligent sensors are connected to the Internet, and the process data transmitted through network communication may be tampered with externally. This type of network attack on the control layer and monitoring layer of the industrial process will directly manipulate the physical quantities in the controlled process, thereby causing damage to the physical entity. The existing process monitoring system can effectively detect anomalies in the controlled system, including spontaneous failures within the system and external data tampering behaviors, and realize online real-time diagnosis and alarm. However, more and more novel external data tampering methods use control theory knowledge and carry out targeted designs when tampering with the input and output data of the controlled process, so that the existing process monitoring methods based on residual generation and residual evaluation cannot detect abnormal states, achieving the purpose of hidden tampering. This type of method causes the actual running trajectory of the system variables to deviate from the expected state trajectory, and the automated monitoring mechanism does not issue an alarm for this abnormal situation. The resulting safety-critical system failure will not only affect the production and manufacturing process and reduce product quality, but also damage the physical equipment entity, causing high-risk accidents such as system loss of control, leakage of hazardous materials, and explosion.

[0003] Therefore, it is very important to understand the design defects of the process monitoring system and the principles of external data tampering. However, there is currently no covert method for regulating the state variables of the industrial control system, which cannot detect the design defects of the process monitoring system and the correlation of external data, resulting in loopholes in the security protection training of the industrial control system, resulting in a lack of corresponding security protection strategies and increasing the risks of industrial control. Summary of the invention

[0004] The present invention aims to solve the problem that the existing data-driven industrial control system does not have a state variable regulation method, which leads to the inability to effectively discover changes in data trajectories, as well as the shortcomings of automated monitoring technology and industrial process fault diagnosis technology.

[0005] A method for regulating state variables of an industrial control system in a device-edge-cloud collaborative manner. The method requires pre-processing of data in an offline phase before regulating state variables of online data of a controlled industrial process. Then, based on the data processed in the offline phase, state variables of the online data of the controlled industrial process are regulated.

[0006] The process of pre-processing the offline data includes:

[0007] S100. Construct the nominal system and residual generator of the controlled process on the cloud server:

[0008] Step A1: Obtain the state space model of the controlled industrial process Among them, u(k) and y(k) are the control input and sensor measurement output of the controlled industrial process, x(k) is the n-dimensional state variable, p(k) is the process noise, and o(k) is the measurement noise; A, B, C, and D are the system matrix, control matrix, output matrix, and direct transfer matrix in the state equation and output equation, respectively;

[0009] Step A2: Obtain the covariance matrix Q of the process noise term p(k) in the state equation and the covariance matrix R of the sensor measurement noise term o(k) in the output equation by a system identification method;

[0010] Step A3: Reconstruct the residual generator r of the output variable y according to the type and order of the residual generator used in the process monitoring system s (k), which is actually a generalized residual vector with a window length of s constructed based on the output variable residual r(k), and the generalized residual sequence r is calculated s The covariance matrix of (k) is denoted as

[0011] Residual Generator Among them, y s (k),u s (k), p s (k), o s (k) is a data set containing s samples, H u,s , H p,s is the generalized input coefficient matrix and the generalized process noise coefficient matrix; v s Represents an equivalent vector, satisfying: v s Γ s =0, where Γ s is the augmented observability matrix;

[0012] S200, calculate the hidden matrix H on the cloud server a :

[0013] Step B1: Set the length of the equivalent space to s and construct the generalized observability matrix

[0014] Step B2: From the matrix Γ s Select a hidden matrix H from the left null space of a =Null(Γ s ), Null is an operator for finding the left zero matrix of a matrix; the equivalent vector v s Through the hidden matrix Ha Construct any row of ;

[0015] S300, calculating the optimal state estimate in the absence of abnormalities on the cloud server:

[0016] Step C1: Based on the equivalent space length, use C and A to construct the generalized process noise coefficient matrix H p,s , using A, B, C and D to generalize the input coefficient matrix H u,s ;

[0017] Step C2: The process noise and measurement noise are expressed as composite noise ζ(k), ζ(k) = H p,s p s (k)+o s (k) and its covariance matrix is ​​denoted as Σ ζ ; Based on the noise covariance matrices Q and R obtained by S100, calculate

[0018] Step C3: Collect online input u(k) and output data y(k), and construct an input vector u with a window length of s s (k), output vector y s (k); Calculate the generalized output vector without abnormality

[0019] Step C4: Calculate the state estimation projection matrix

[0020] Step C5: Use weighted least squares method to calculate the optimal state estimate without abnormality

[0021] The process of regulating the state variables of the online data of the controlled industrial process based on the data processed in the offline stage includes the following steps:

[0022] S400, setting the industrial control security drill strategy; if the state offset strategy is adopted, then enter S500; if the state variable control strategy is adopted, then enter S600;

[0023] S500: Generate a data sequence a that maximizes the state trajectory deviation on the edge node s (k); The specific steps include:

[0024] Step E1: Set the upper limit of the control energy value, denoted as M a ;

[0025] Step E2: Read the hidden matrix H calculated in the offline phase a and the state estimation projection matrix

[0026] Step E3: Calculate the matrix The principal characteristic vector of , that is, the eigenvector corresponding to the maximum eigenvalue γ * (k);

[0027] Step E4: Calculate the optimal sequence of state shifts in represents the normalized attack vector, M a To adjust the upper limit of the energy value, enter S700;

[0028] S600, generating a state variable control data sequence on the edge node:

[0029] Step F1: Set the virtual state trajectory to f(k);

[0030] Step F2: Collect online control command input u(k) and sensor measurement output data y(k), and construct a stacked control command input vector u with a window length of s s (k), stacked sensor measurement output vector y s (k); Calculate the generalized output vector

[0031] Step F3: Generate state variable control data sequence,

[0032] S700, implement data injection on the terminal collection and injection system:

[0033] Step G1: From Extract the subsequence from the sequence, and assign a(k-s+1), a(k-s+2)…a(k) to the subsequence a in a cyclical manner. s,1 (k), a s,2 (k)…a s,m (k);

[0034] Step G2: intercepting sensor transmission data in the network communication channel and unpacking it;

[0035] Step G3: Subsequence a s,1 (k), a s,2 (k)…a s,m Each element a in (k) i (k) and the measured value y in the i-th sensor channel in turn i (k) Add and calculate the tampered data y a,i (k) = y i (k)+a i (k);

[0036] Step G4: Repackage the tampered data and resend it through the network communication channel.

[0037] Furthermore, the generalized process noise coefficient matrix H in step C1 is p,s and the generalized input coefficient matrix H u,s as follows:

[0038]

[0039]

[0040] Furthermore, the input vector constructed in step C3 Constructed output vector

[0041] Furthermore, the stack control instruction input vector constructed in step F2 is Constructed stacked sensor measurement output vector

[0042] Furthermore, in step G1, a(k-s+1), a(k-s+2)…a(k) are periodically assigned to the subsequence a. s,1 (k), a s,2 (k)…a s,m The process of (k) is as follows:

[0043] Assign a(k-s+1), a(k-s+2)…a(k-s+m) to a in sequence s,1 (k), a s,2 (k)…a s,m (k), assign a(k-s+m+1), a(k-s+m+2)…a(k-s+2m) to a in sequence s,1 (k), a s,2 (k)…a s,m (k), and follow this rule until a(k-m+1), a(k-m+2)…a(k) are assigned to a in sequence s,1 (k), a s,2 (k)…a s,m (k);

[0044] subsequence a s,i (k)∈R L×1 , Where m is the dimension of the output.

[0045] A method for regulating state variables of an industrial control system in a device-edge-cloud collaborative manner. The method requires pre-processing of data in an offline phase before regulating state variables of online data of a controlled industrial process. Then, based on the data processed in the offline phase, state variables of the online data of the controlled industrial process are regulated.

[0046] The process of pre-processing the offline data includes:

[0047] S100, construct the nominal system and residual generator of the controlled process:

[0048] Step A1: Obtain the state space model of the controlled industrial process Among them, u(k) and y(k) are the control input and sensor measurement output of the controlled industrial process, x(k) is the n-dimensional state variable, p(k) is the process noise, and o(k) is the measurement noise; A, B, C, and D are the system matrix, control matrix, output matrix, and direct transfer matrix in the state equation and output equation, respectively;

[0049] Step A3: Reconstruct the residual generator r of the output variable y according to the type and order of the residual generator used in the process monitoring system s (k), which is actually a generalized residual vector with a window length of s constructed based on the output variable residual r(k), and the generalized residual sequence r is calculated s The covariance matrix of (k) is denoted as

[0050] Residual Generator Among them, y s (k),u s (k), p s (k), o s (k) is a data set containing s samples, H u,s , H p,s is the generalized input coefficient matrix and the generalized process noise coefficient matrix; v s Represents an equivalent vector, satisfying: v s Γ s =0, where Γ s is the augmented observability matrix;

[0051] S200, calculate the hidden matrix H on the cloud server a :

[0052] Step B1: Set the length of the equivalent space to s and construct the generalized observability matrix

[0053] Step B2: From the matrix Γ s Select a hidden matrix H from the left null space of a=Null(Γ s ), Null is an operator for finding the left zero matrix of a matrix; the equivalent vector v s Through the hidden matrix H a Construct any row of ;

[0054] S300, calculating the optimal state estimate in the absence of abnormalities on the cloud server:

[0055] Step C1: Based on the equivalent space length, use C and A to construct the generalized process noise coefficient matrix H p,s , using A, B, C and D to generalize the input coefficient matrix H u,s ;

[0056] Step C2: The process noise and measurement noise are expressed as composite noise ζ(k), ζ(k) = H p,s p s (k)+o s (k) and its covariance matrix is ​​denoted as Σ ζ ; in, Yes s (k) the covariance matrix; ν s is the equivalent vector constructed in step B2;

[0057] Step C3: Collect online input u(k) and output data y(k), and construct an input vector u with a window length of s s (k), output vector y s (k); Calculate the generalized output vector without abnormality

[0058] Step C4: Calculate the state estimation projection matrix

[0059] Step C5: Use weighted least squares method to calculate the optimal state estimate without abnormality

[0060] The process of regulating the state variables of the online data of the controlled industrial process based on the data processed in the offline stage includes the following steps:

[0061] S400, setting the industrial control security drill strategy; if the state offset strategy is adopted, then enter S500; if the state variable control strategy is adopted, then enter S600;

[0062] S500: Generate a data sequence a that maximizes the state trajectory deviation on the edge node s (k); The specific steps include:

[0063] Step E1: Set the upper limit of the control energy value, denoted as M a ;

[0064] Step E2: Read the hidden matrix H calculated in the offline phase a and the state estimation projection matrix

[0065] Step E3: Calculate the matrix The principal characteristic vector of , that is, the eigenvector corresponding to the maximum eigenvalue γ * (k);

[0066] Step E4: Calculate the optimal sequence of state shifts in represents the normalized attack vector, M a To adjust the upper limit of the energy value, enter S700;

[0067] S600, generating a state variable control data sequence on the edge node:

[0068] Step F1: Set the virtual state trajectory to f(k);

[0069] Step F2: Collect online control command input u(k) and sensor measurement output data y(k), and construct a stacked control command input vector u with a window length of s s (k), stacked sensor measurement output vector y s (k)

[0070]

[0071] Compute the generalized output vector

[0072] Step F3: Generate state variable control data sequence,

[0073] S700, implement data injection on the terminal collection and injection system:

[0074] Step G1: From Extract the subsequence from the sequence, and assign a(k-s+1), a(k-s+2)…a(k) to the subsequence a in a cyclical manner. s,1 (k), a s,2 (k)…a s,m (k);

[0075] Step G2: intercepting sensor transmission data in the network communication channel and unpacking it;

[0076] Step G3: Subsequence a s,1 (k), a s,2 (k)…a s,mEach element a in (k) i (k) and the measured value y in the i-th sensor channel in turn i (k) Add and calculate the tampered data y a,i (k) = y i (k)+a i (k);

[0077] Step G4: Repackage the tampered data and resend it through the network communication channel.

[0078] Furthermore, the generalized process noise coefficient matrix H in step C1 is p,s and the generalized input coefficient matrix H u,s as follows:

[0079]

[0080]

[0081] Furthermore, the input vector constructed in step C3 Constructed output vector

[0082] Furthermore, the stack control instruction input vector constructed in step F2 is Constructed stacked sensor measurement output vector

[0083] Furthermore, in step G1, a(k-s+1), a(k-s+2)…a(k) are periodically assigned to the subsequence a. s,1 (k), a s,2 (k)…a s,m The process of (k) is as follows:

[0084] Assign a(k-s+1), a(k-s+2)…a(k-s+m) to a in sequence s,1 (k), a s,2 (k)…a s,m (k), assign a(k-s+m+1), a(k-s+m+2)…a(k-s+2m) to a in sequence s,1 (k), a s,2 (k)…a s,m (k), and follow this rule until a(k-m+1), a(k-m+2)…a(k) are assigned to a in sequence s,1 (k), a s,2 (k)…a s,m (k);

[0085] subsequence a s,i (k)∈RL×1 , Where m is the dimension of the output.

[0086] The present invention has the following advantages:

[0087] 1. The method provided by the present invention can make the state of the controlled system deviate from the ideal state trajectory to the greatest extent under the constraint of limited energy.

[0088] 2. The present invention provides a state variable control method, which can provide a powerful adversarial training opponent for industrial control system attack and defense drills, thereby solving the problem of not being able to effectively discover changes in data trajectories, and then can keep the existing anomaly detection mechanism hidden while regulating the system state, making it impossible for the process monitoring system to detect anomalies.

[0089] 3. The device required for deploying the method of the present invention independently designs the storage, computing and communication functions, and the end-edge-cloud subsystems cooperate with each other, which can easily meet the performance requirements of real-time regulation of the state of the dynamically controlled system. BRIEF DESCRIPTION OF THE DRAWINGS

[0090] Figure 1 It is a flowchart of the method proposed in the specific implementation method.

[0091] Figure 2 It is a functional structure diagram of the device required for deploying the method of the present invention.

[0092] Figure 3 It is the change of system state before and after injecting the optimal data tampering sequence.

[0093] Figure 4 It is the change of system state before and after injecting non-optimal data tampering sequence. DETAILED DESCRIPTION

[0094] The present invention proposes a method for adjusting the state variables of an industrial control system and a functional structure of a device for deploying and implementing the method, aiming to improve the deficiencies in the prior art in the security protection training of industrial control systems. The method can be used for attack and defense drills in a digital shooting range environment, and can help improve the security protection capabilities of industrial control systems. Aiming at the deficiencies of existing data-driven industrial control system automation monitoring technology and industrial process fault diagnosis technology, the present invention can make the state of the controlled system deviate from the ideal state trajectory to the greatest extent under limited energy constraints, or force the state of the controlled system to change according to a preset virtual trajectory, providing a powerful confrontation training scenario for the attack and defense drills of the industrial control system. The device of the deployment method provided by the present invention is composed of three parts: a terminal acquisition injection system, an edge computing node system, and a cloud server, and the real-time regulation of the state variables of the industrial system is achieved through the cooperation of the three.

[0095] Specific implementation method 1: Combination and Figure 1 and Figure 2 To explain this embodiment,

[0096] This embodiment is a method and device for regulating the state variables of an industrial control system in an end-edge-cloud collaborative manner, which provides a method for regulating the state variables of an industrial control system, which can make the state of the controlled system deviate from the ideal state trajectory to the greatest extent under the constraint of limited energy, or force the state of the controlled system to change according to a preset virtual trajectory, while preventing the original process monitoring system from sounding an alarm. At the same time, a device for deploying and implementing this method is provided, which includes three parts: a terminal acquisition injection system, an edge computing node, and a cloud server. In order to efficiently and reasonably allocate functions and reduce hardware requirements, the three systems are divided into storage modules (or cache modules), computing modules, and communication modules according to their functions. It is mainly implemented through the following technical solutions:

[0097] Terminal acquisition and injection system: mainly used to collect online data and inject state-controlled data sequences. Specifically, the cache module of the terminal acquisition and injection system is used to temporarily store data collected from the sensor network of the industrial control system and edge node data. Online data includes control input data of the controlled industrial process and sensor network data;

[0098] The first communication module of the terminal acquisition injection system is used to send original data to the edge node and receive the tampered data of the edge node; the original data is the online data before being sent to the edge node (the online data before tampering);

[0099] The second communication module of the terminal acquisition injection system is used to collect online data and send tampered data to the sensor network;

[0100] Edge nodes: Mainly used for real-time data processing and small-scale real-time computing. Specifically,

[0101] Communication module of edge node: used to receive data from cloud servers and data sent by terminal collection and injection systems, and also used to send data calculated by the calculation module to the terminal collection and injection system (actually tampered data).

[0102] Storage module of edge node: used to store data sent by the terminal collection injection system and data sent by the cloud server;

[0103] The computing module of the edge node is used to calculate the optimal tampering data sequence of state offset and the state control data sequence;

[0104] Cloud server: mainly used to store historical data and system models, as well as offline calculation of complex formulas and large matrices. Specifically, the storage module of the cloud server is mainly responsible for storing the nominal model of the controlled industrial process, the residual generator and the newly collected data; historical data is pre-collected offline data, and offline data includes control input data and sensor network data of the controlled industrial process;

[0105] The computing module of the cloud server is mainly used to calculate the concealment matrix, composite noise covariance matrix and state estimation projection matrix offline, and calculate the main eigenvector online;

[0106] Communication module of the cloud server: mainly used to send state variable control instructions and parameters, offline calculated matrices and online calculated vectors, and receive newly collected process data.

[0107] The algorithms running on the cloud server and edge nodes cooperate with each other to achieve real-time control of industrial system state variables. Figure 1 As shown, the functional structure diagram of the device required for deploying the method of the present invention is as follows Figure 2 shown.

[0108] The end-edge-cloud collaborative industrial control system state variable regulation method described in the present invention is mainly implemented through the following steps. It should be noted that the first to third steps belong to the offline design stage, and the fourth to seventh steps belong to the online deployment stage.

[0109] Offline design phase

[0110] The first step is to build the nominal system and residual generator of the controlled process on the cloud server. The specific steps include:

[0111] Step A1: Obtain the state space model of the controlled industrial process

[0112]

[0113] Where u(k)∈R l y(k)∈R m are the control input and sensor measurement output of the controlled industrial process, respectively, x(k)∈R n is the n-dimensional state variable, p(k)∈R n is the process noise, o(k)∈R m is the measurement noise. A, B, C, and D are the system matrix, control matrix, output matrix, and direct transfer matrix in the state equation and output equation, respectively.

[0114] Step A2: Obtain the covariance matrix Q of the process noise term p(k) in the state equation and the covariance matrix R of the sensor measurement noise term o(k) in the output equation by a system identification method.

[0115] The process noise p(k) and the measurement noise o(k) are independent Gaussian white noises and satisfy p(k)~N(0,Q), o(k)~N(0,R), where the covariance matrix Q≥0, R≥0.

[0116] If the information about the process noise term and the sensor measurement noise term is incomplete, this step can be skipped.

[0117] Step A3: According to the process monitoring system (a system for monitoring the controlled industrial process, such as Figure 2 The type and order of the residual generator used to reconstruct the output variable y is shown in Figure 1. s (k), which is actually a generalized residual vector with a window length of s constructed based on the output variable residual r(k), and the generalized residual sequence r is calculated s The covariance matrix of (k) is denoted as

[0118]

[0119] Residual Generator s (k):

[0120]

[0121] Among them, y s (k),u s (k), p s (k), o s (k) is a data set containing s samples, H u,s , H p,s are the generalized input coefficient matrix and the generalized process noise coefficient matrix;

[0122] v s Represents an equivalent vector, satisfying: v s Γ s =0, recorded as Among them, Γ s is the augmented observability matrix, Represents Γ s The left null space of .

[0123] Step 2: Calculate the hidden matrix H on the cloud server a :

[0124] Step B1: Construct the generalized observability matrix Γ s, where the length of the parity space is set to s (the parity space is a vector space composed of feature vectors under different fault modes of the system, which is used to represent all possible fault states);

[0125]

[0126] Step B2: From the matrix Γ s Select a hidden matrix H from the left null space of a =Null(Γ s ), Null is an operator for finding the left zero matrix of a matrix; the equivalent vector v s Through the hidden matrix H a Construct any row of .

[0127] Step 3: Calculate the optimal estimate of the state without abnormal conditions on the cloud server:

[0128] Step C1: Construct the generalized process noise coefficient matrix H p,s and the generalized input coefficient matrix H u,s :

[0129]

[0130] Step C2: To simplify the notation, the process noise and the measurement noise are considered together and expressed as composite noise ζ(k), ζ(k) = H p,s p s (k)+o s (k) and its covariance matrix is ​​denoted as Σ ζ ; Read the noise covariance matrices Q and R obtained in the first step and calculate

[0131] If there is incomplete information about the noise covariance, that is, Q or R is unknown, then in, Yes s (k) the covariance matrix; ν s is the equivalent vector constructed in step B2. In fact, even if the information is complete, it is still possible to use of.

[0132] To standardize the use of process data, is a data vector, introducing the following notation:

[0133]

[0134] Where: s is the window length of fault detection, N is the capacity of the data set when collecting data, both are integers, ω(k) can be x(k), u(k), y(k) and noise p(k) and o(k), all of which have similar forms, and ξ is the corresponding data dimension; Represents the real number space.

[0135] Step C3: Collect online input u(k) and output data y(k), and construct input and output vectors with a window length of s:

[0136]

[0137] Compute the generalized output vector for the non-anomalous case

[0138] Step C4: Calculate the state estimation projection matrix

[0139] Step C5: Use weighted least squares method to calculate the optimal state estimate without abnormality

[0140] Online deployment phase

[0141] Step 4: Set the industrial control security drill strategy. If the state offset strategy is adopted, go to step 5; if the state variable control strategy is adopted, go to step 6.

[0142] Step 5: Generate a data sequence that maximizes the state trajectory deviation (i.e., the optimal state deviation sequence) on the edge node s (k). The specific steps include:

[0143] Step E1: Set the upper limit of the control energy value, denoted as M a .

[0144] Step E2: Read the hidden matrix H calculated in the offline phase a and the state estimation projection matrix

[0145] Step E3: Calculate the matrix The principal characteristic vector of , that is, the eigenvector corresponding to the maximum eigenvalue γ * (k).

[0146] Step E4: Calculate the optimal sequence of state shifts in represents the normalized attack vector, M a To regulate the upper limit of the energy value, proceed to step 7.

[0147] Step 6: Generate state variable control data sequence on edge nodes:

[0148] Step F1: Set the virtual state trajectory to f(k).

[0149] Step F2: Collect online control command input u(k) and sensor measurement output data y(k), and construct a stacked control command input vector u with a window length of s s (k), stacked sensor measurement output vector y s (k)

[0150]

[0151] Compute the generalized output vector

[0152] Step F3: Generate state variable control data sequence,

[0153] Step 7: Implement data injection on the terminal collection and injection system:

[0154] Step G1: From Extract the subsequence from the sequence, and assign a(k-s+1), a(k-s+2)…a(k) to the subsequence a in a cyclical manner. s,1 (k), a s,2 (k)…a s,m (k), that is, assign a(k-s+1), a(k-s+2)…a(k-s+m) to a in sequence s,1 (k), a s,2 (k)…a s,m (k), assign a(k-s+m+1), a(k-s+m+2)…a(k-s+2m) to a in sequence s,1 (k), a s,2 (k)…a s,m (k), and follow this rule until a(k-m+1), a(k-m+2)…a(k) are assigned to a in sequence s,1 (k), a s,2 (k)…a s,m (k), subsequence a s,i (k)∈R L×1 , Where m is the dimension of the output.

[0155] Step G2: intercepting sensor transmission data in the network communication channel and unpacking it.

[0156] Step G3: Subsequence a s,1 (k), a s,2 (k)…a s,m Each element a in (k)i (k) and the measured value y in the i-th sensor channel in turn i (k) Add and calculate the tampered data y a,i (k) = y i (k)+a i (k).

[0157] Step G4: Repackage the tampered data and resend it through the network communication channel.

[0158] Example

[0159] The controlled industrial system in this embodiment is a liquid level control system, whose prototype is the DTS200 experimental system, which is a typical regulation problem in process industry and process control. The state variable x of the system is the liquid level height of three containers (water tanks), which is expected to be maintained at 45cm, 15cm, and 30cm by controller adjustment. The control input variable u of the system is the water injection flow rate. A pressure sensor is equipped at the bottom of each container, which can calculate the corresponding liquid level y in real time.

[0160] The data measured by the pressure sensor is transmitted to the dSPACE control system simulation platform through the data acquisition card and then interacts with the PC to realize terminal data acquisition and data injection on the PC side. The sampling frequency is set to 0.5Hz. 2000 (4000 seconds) sample data under steady-state operation of the acquisition system are used for the offline design stage. In the online deployment stage, 3000 (6000 seconds) sample data are generated and state variable regulation experiments are performed on them. The state offset strategy is adopted for the samples from the 2000th second to the 5300th second, which can be regarded as a special case of state variable regulation. In the following steps, we focus on the dimensions of the calculated matrices and vectors, because they directly reflect the feasibility of the method and the computing and storage requirements that need to be considered during deployment.

[0161] Offline design phase

[0162] The first step is to build the nominal system and residual generator of the controlled process on the cloud server.

[0163] Step A1: Identify the system and obtain the system matrix

[0164]

[0165] D=0.

[0166] Step A2: The information about the noise covariance is incomplete, so this step is skipped.

[0167] Step A3: The master process monitoring system adopts an anomaly detection algorithm based on equivalent space to calculate the covariance matrix of the generalized residual sequence, which is a matrix with 12 rows and 12 columns.

[0168] Step 2: Calculate the concealment matrix on the cloud server.

[0169] Step B1: The length of the equivalent space is set to 4. Construct the generalized observability matrix Γ s , is a matrix with 12 rows and 3 columns.

[0170] Step B2: From the matrix Γ s A hidden matrix is ​​selected from the left null space, which is a matrix with 12 rows and 12 columns.

[0171] The third step is to calculate the optimal estimate of the state under normal circumstances on the cloud server.

[0172] Step C1: Construct the generalized process noise coefficient matrix H p,s , is a 12-row by 12-column matrix; and the generalized input coefficient matrix H u,s , is a matrix with 12 rows and 8 columns.

[0173] Step C2: Select the equivalent vector according to the hidden matrix:

[0174] ν s =10 4 *

[0175] [-2.17 -3.04 4.28 -85.35 -83.45 168.84 2.17 3.04 -4.28 -0.06 -0.100.12], and estimate the covariance matrix Σ of the composite noise term ζ , is a matrix with 12 rows and 12 columns.

[0176] Step C3: Collect online input and output data u(k), y(k), and construct input and output vectors u with a window length of s s (k), y s (k) and calculate the generalized output vector in the absence of anomalies.

[0177] Step C4: Calculate the state estimation projection matrix It is a matrix with 3 rows and 12 columns.

[0178] Step C5: The state estimation value may be calculated as needed.

[0179] Online deployment phase

[0180] The fourth step is to adopt a state shift strategy (the state shift strategy can also be regarded as a special case of state variable regulation).

[0181] Step 5: Generate a data sequence on the edge node that maximizes the state trajectory deviation.

[0182] Step E1: Set the upper limit M of the control energy value a =10.

[0183] Step E2: Read the hidden matrix and state estimation projection matrix calculated in the offline stage.

[0184] Step E3: Calculate the principal characteristic vector γ * (k) is a 12-dimensional column vector.

[0185] Step E4: Calculate the optimal sequence a of state offset s (k):

[0186] a s (k)=[0.35 0.09 0.38 0.40 0.15 0.13 0.35 0.09 0.37 0.44 0.21 -0.06]

[0187] Then proceed to step seven.

[0188] Step 7: Implement data injection on the terminal collection and injection system.

[0189] Step G1: From a s (k) Extract subsequence:

[0190] a s,1 (k) = [0.35 0.40 0.35 0.44]

[0191] a s,2 (k) = [0.09 0.15 0.09 0.21]

[0192] a s,3 (k) = [0.38 0.13 0.37 -0.06] Step G2: intercept the sensor transmission data in the network communication channel and unpack it.

[0193] Step G3: Calculate and obtain the tampered data.

[0194] Step G4: Repackage the tampered data and resend it through the network communication channel.

[0195] According to the above steps and the method of the present invention, the optimal data sequence is injected, and the system state changes as shown in the attached figure. Figure 3 As shown in the figure, it can be seen that for the sample data under the state deviation strategy, the three liquid level heights have all dropped significantly, achieving the goal of making the system state deviate from the expected trajectory. Figure 3 The figure shows the change of system state when the method of the present invention is not used but only an arbitrary non-optimized data sequence is injected: although the liquid level height of the controlled system also changes, the system state does not significantly deviate from the expected trajectory. Figure 3 and attached Figure 4 The curve can illustrate the effectiveness and superiority of the method of the present invention.

[0196] The above calculation examples of the present invention are only used to explain the calculation model and calculation process of the present invention in detail, and are not intended to limit the implementation methods of the present invention. For ordinary technicians in the relevant field, other different forms of changes or modifications can be made based on the above description. It is impossible to list all the implementation methods here. All obvious changes or modifications derived from the technical solution of the present invention are still within the scope of protection of the present invention.

Claims

1. A method for regulating state variables of an industrial control system in a device-edge-cloud collaborative manner, characterized in that: Before the method performs state variable regulation on the online data of the controlled industrial process, it is necessary to process the data in the offline stage in advance, and then perform state variable regulation on the online data of the controlled industrial process based on the data processed in the offline stage; The process of pre-processing the offline data includes: S100. Construct the nominal system and residual generator of the controlled process on the cloud server: Step A1: Obtain the state space model of the controlled industrial process Among them, u(k) and y(k) are the control input and sensor measurement output of the controlled industrial process, x(k) is the n-dimensional state variable, p(k) is the process noise, and o(k) is the measurement noise; A, B, C, and D are the system matrix, control matrix, output matrix, and direct transfer matrix in the state equation and output equation, respectively; Step A2: Obtain the covariance matrix Q of the process noise term p(k) in the state equation and the covariance matrix R of the sensor measurement noise term o(k) in the output equation by a system identification method; Step A3: Reconstruct the residual generator r of the output variable y according to the type and order of the residual generator used in the process monitoring system s (k), which is actually a generalized residual vector with a window length of s constructed based on the output variable residual r(k), and the generalized residual sequence r is calculated s The covariance matrix of (k) is denoted as Residual Generator Among them, y s (k),u s (k), p s (k), o s (k) is a data set containing s samples, H u,s , H p,s is the generalized input coefficient matrix and the generalized process noise coefficient matrix; v s Represents an equivalent vector, satisfying: v s Γ s =0, where Γ s is the augmented observability matrix; S200, calculate the hidden matrix H on the cloud server a : Step B1: Set the length of the equivalent space to s and construct the generalized observability matrix Step B2: From the matrix Γ s Select a hidden matrix H from the left null space of a =Null(Γ s ), Null is an operator for finding the left zero matrix of a matrix; the equivalent vector v s Through the hidden matrix H a Construct any row of ; S300, calculating the optimal state estimate in the absence of abnormalities on the cloud server: Step C1: Based on the equivalent space length, use C and A to construct the generalized process noise coefficient matrix H p,s , using A, B, C and D to generalize the input coefficient matrix H u,s ; Step C2: The process noise and measurement noise are expressed as composite noise ζ(k), ζ(k) = H p,s p s (k)+o s (k) and its covariance matrix is ​​denoted as Σ ζ ; Based on the noise covariance matrices Q and R obtained by S100, calculate Step C3: Collect online input u(k) and output data y(k), and construct an input vector u with a window length of s s (k), output vector y s (k); Calculate the generalized output vector without abnormality Step C4: Calculate the state estimation projection matrix Step C5: Use weighted least squares method to calculate the optimal state estimate without abnormality The process of regulating the state variables of the online data of the controlled industrial process based on the data processed in the offline stage includes the following steps: S400, setting the industrial control security drill strategy; if the state offset strategy is adopted, then enter S500; if the state variable control strategy is adopted, then enter S600; S500: Generate a data sequence a that maximizes the state trajectory deviation on the edge node s (k); The specific steps include: Step E1: Set the upper limit of the control energy value, denoted as M a ; Step E2: Read the hidden matrix H calculated in the offline phase a and the state estimation projection matrix Step E3: Calculate the matrix The principal characteristic vector of , that is, the eigenvector corresponding to the maximum eigenvalue γ * (k); Step E4: Calculate the optimal sequence of state shifts in represents the normalized attack vector, M a To adjust the upper limit of the energy value, enter S700; S600, generating a state variable control data sequence on the edge node: Step F1: Set the virtual state trajectory to f(k); Step F2: Collect online control command input u(k) and sensor measurement output data y(k), and construct a stacked control command input vector u with a window length of s s (k), stacked sensor measurement output vector y s (k); Calculate the generalized output vector Step F3: Generate state variable control data sequence, S700, implement data injection on the terminal collection and injection system: Step G1: From Extract the subsequence from the sequence, and assign a(k-s+1), a(k-s+2)…a(k) to the subsequence a in a cyclical manner. s,1 (k), a s,2 (k)…a s,m (k); Step G2: intercepting sensor transmission data in the network communication channel and unpacking it; Step G3: Subsequence a s,1 (k), a s,2 (k)…a s,m Each element a in (k) i (k) and the measured value y in the i-th sensor channel in turn i (k) Add and calculate the tampered data y a,i (k) = y i (k)+a i (k); Step G4: Repackage the tampered data and resend it through the network communication channel.

2. According to the end-edge-cloud collaborative industrial control system state variable control method of claim 1, it is characterized in that: The generalized process noise coefficient matrix H in step C1 p,s and the generalized input coefficient matrix H u,s as follows:

3. According to the end-edge-cloud collaborative industrial control system state variable control method of claim 1, it is characterized in that: The input vector constructed in step C3 Constructed output vector 4. According to the end-edge-cloud collaborative industrial control system state variable control method of claim 1, it is characterized in that: The stack control instruction input vector constructed in step F2 Constructed stacked sensor measurement output vector 5. A method for controlling state variables of an industrial control system in a device-edge-cloud collaboration according to any one of claims 1 to 4, characterized in that: In step G1, a(k-s+1), a(k-s+2)…a(k) are periodically assigned to the subsequence a. s,1 (k), a s,2 (k)…a s,m The process of (k) is as follows: Assign a(k-s+1), a(k-s+2)…a(k-s+m) to a in sequence s,1 (k), a s,2 (k)…a s,m (k), assign a(k-s+m+1), a(k-s+m+2)…a(k-s+2m) to a in sequence s,1 (k), a s,2 (k)…a s,m (k), and follow this rule until a(k-m+1), a(k-m+2)…a(k) are assigned to a in sequence s,1 (k), a s,2 (k)…a s,m (k); subsequence Where m is the dimension of the output.

6. A method for regulating state variables of an industrial control system in a device-edge-cloud collaborative manner, characterized in that: Before the method performs state variable regulation on the online data of the controlled industrial process, it is necessary to process the data in the offline stage in advance, and then perform state variable regulation on the online data of the controlled industrial process based on the data processed in the offline stage; The process of pre-processing the offline data includes: S100, construct the nominal system and residual generator of the controlled process: Step A1: Obtain the state space model of the controlled industrial process Among them, u(k) and y(k) are the control input and sensor measurement output of the controlled industrial process, x(k) is the n-dimensional state variable, p(k) is the process noise, and o(k) is the measurement noise; A, B, C, and D are the system matrix, control matrix, output matrix, and direct transfer matrix in the state equation and output equation, respectively; Step A3: Reconstruct the residual generator r of the output variable y according to the type and order of the residual generator used in the process monitoring system s (k), which is actually a generalized residual vector with a window length of s constructed based on the output variable residual r(k), and the generalized residual sequence r is calculated s The covariance matrix of (k) is denoted as Residual Generator Among them, y s (k),u s (k), p s (k), o s (k) is a data set containing s samples, H u,s , H p,s is the generalized input coefficient matrix and the generalized process noise coefficient matrix; v s Represents an equivalent vector, satisfying: v s Γ s =0, where Γ s is the augmented observability matrix; S200, calculate the hidden matrix H on the cloud server a : Step B1: Set the length of the equivalent space to s and construct the generalized observability matrix Step B2: From the matrix Γ s Select a hidden matrix H from the left null space of a =Null(Γ s ), Null is an operator for finding the left zero matrix of a matrix; the equivalent vector v s Through the hidden matrix H a Construct any row of ; S300, calculating the optimal state estimate in the absence of abnormalities on the cloud server: Step C1: Based on the equivalent space length, use C and A to construct the generalized process noise coefficient matrix H p,s , using A, B, C and D to generalize the input coefficient matrix H u,s ; Step C2: The process noise and measurement noise are expressed as composite noise ζ(k), ζ(k) = H p,s p s (k)+o s (k) and its covariance matrix is ​​denoted as Σ ζ ; Among them, Σ rs Yes s (k) the covariance matrix; ν s is the equivalent vector constructed in step B2; Step C3: Collect online input u(k) and output data y(k), and construct an input vector u with a window length of s s (k), output vector y s (k); Calculate the generalized output vector without abnormality Step C4: Calculate the state estimation projection matrix Step C5: Use weighted least squares method to calculate the optimal state estimate without abnormality The process of regulating the state variables of the online data of the controlled industrial process based on the data processed in the offline stage includes the following steps: S400, setting the industrial control security drill strategy; if the state offset strategy is adopted, then enter S500; if the state variable control strategy is adopted, then enter S600; S500: Generate a data sequence a that maximizes the state trajectory deviation on the edge node s (k); The specific steps include: Step E1: Set the upper limit of the control energy value, denoted as M a ; Step E2: Read the hidden matrix H calculated in the offline phase a and the state estimation projection matrix Step E3: Calculate the matrix The principal characteristic vector of , that is, the eigenvector corresponding to the maximum eigenvalue γ * (k); Step E4: Calculate the optimal sequence of state shifts in represents the normalized attack vector, M a To adjust the upper limit of the energy value, enter S700; S600, generating a state variable control data sequence on the edge node: Step F1: Set the virtual state trajectory to f(k); Step F2: Collect online control command input u(k) and sensor measurement output data y(k), and construct a stacked control command input vector u with a window length of s s (k), stacked sensor measurement output vector y s (k) Compute the generalized output vector Step F3: Generate state variable control data sequence, S700, implement data injection on the terminal collection and injection system: Step G1: From Extract the subsequence from the sequence, and assign a(k-s+1), a(k-s+2)…a(k) to the subsequence a in a cyclical manner. s,1 (k), a s,2 (k)…a s,m (k); Step G2: intercepting sensor transmission data in the network communication channel and unpacking it; Step G3: Subsequence a s,1 (k), a s,2 (k)…a s,m Each element a in (k) i (k) and the measured value y in the i-th sensor channel in turn i (k) Add and calculate the tampered data y a,i (k) = y i (k)+a i (k); Step G4: Repackage the tampered data and resend it through the network communication channel.

7. The method for controlling state variables of an industrial control system in a device-edge-cloud collaboration according to claim 6 is characterized in that: The generalized process noise coefficient matrix H in step C1 p,s and the generalized input coefficient matrix H u,s as follows:

8. The method for controlling state variables of an industrial control system in device-edge-cloud collaboration according to claim 6 is characterized in that: The input vector constructed in step C3 Constructed output vector 9. The method for controlling state variables of an industrial control system in device-edge-cloud collaboration according to claim 6 is characterized in that: The stack control instruction input vector constructed in step F2 Constructed stacked sensor measurement output vector 10. A method for controlling state variables of an industrial control system in device-edge-cloud collaboration according to any one of claims 6 to 9, characterized in that: In step G1, a(k-s+1), a(k-s+2)…a(k) are periodically assigned to the subsequence a. s,1 (k), a s,2 (k)…a s,m The process of (k) is as follows: Assign a(k-s+1), a(k-s+2)…a(k-s+m) to a in sequence s,1 (k), a s,2 (k)…a s,m (k), assign a(k-s+m+1), a(k-s+m+2)…a(k-s+2m) to a in sequence s,1 (k), a s,2 (k)…a s,m (k), and follow this rule until a(k-m+1), a(k-m+2)…a(k) are assigned to a in sequence s,1 (k), a s,2 (k)…a s,m (k); subsequence a s,i (k)∈R L×1 , Where m is the dimension of the output.

Citation Information

Patent Citations

  • Cloud network end cooperative defense method and system based on end-side edge computing

    CN112769796A

  • Multi-working-condition industrial robot motion quality anomaly detection system and method

    CN117793163A

  • Flow battery operation and maintenance system and method based on cloud digital twinning

    CN118101732A

  • Ore dressing multi-production-index optimization decision making system and method based on cloud and mobile terminal

    WO2016110165A1