Industrial control system state estimation method based on coding and decoding protocol

Abstract

The invention discloses an industrial control system state estimation method based on a coding and decoding protocol, and relates to the technical field of coding and decoding protocols and double time scales. Comprising the steps of establishing a dual-time-scale industrial control system model based on a coding and decoding protocol, designing a state estimation strategy based on a singular perturbation system, further introducing an index final bounded condition according to a matrix established in a Lyapunov function, ensuring that an error is attenuated at an index rate, avoiding infinite increase of the error, and improving the accuracy of the error. By optimizing the gain matrix and the error attenuation rate, rapid convergence of errors is achieved, the robustness of the system is enhanced, meanwhile, stability conditions are solved by converting into linear matrix inequality, it is ensured that the system stably operates under quantization errors and disturbance, and through accurate design and optimization, the robustness of the system is improved. The state estimation precision, stability and robustness of the industrial control system are improved, and uncertainty and noise can be effectively dealt with in a complex environment.

Description

Technical Field

[0001] The present invention relates to the technical fields of encoding and decoding protocols and dual-time scale technology, and particularly relates to a method for estimating the state of an industrial control system based on an encoding and decoding protocol. Background Art

[0002] In an industrial control system, there are many links that form fast and slow subsystems, which can be modeled as a dual-time scale singularly perturbed system. For example, in a DC motor, the change speeds of voltage and current are relatively fast, while the change speed of rotational speed is relatively slow. Traditional linear control theory cannot effectively handle such a system with multiple time scales. The emergence of singular perturbation theory fills this gap. In the mathematical model of a permanent magnet synchronous motor, a relatively small inductance is used as the singular perturbation parameter, and the circuit system and the mechanical mechanism are decomposed into fast and slow subsystems, and the system is modeled by coupling these two parts.

[0003] The existing methods for estimating the state of an industrial control system have at least the following technical problems: 1. The existing methods for estimating the state of an industrial control system lack the convergence analysis of quantization errors and the evaluation of parameter coefficients, which is not conducive to the optimization of the estimator performance and the guarantee of system stability. When the system faces quantization errors, noise and other uncertainty factors, it cannot effectively converge and maintain accurate state estimation. In addition, the lack of convergence analysis of quantization errors is not conducive to judging whether the system can work stably in practical applications, which will lead to the inability to effectively control errors in long-term operation, resulting in inaccurate state estimation and a decline in system performance.

[0004] 2. Most of the existing methods for estimating the state of an industrial control system lack a detailed evaluation of the gain matrix, error decay rate and other parameters, which is not conducive to optimizing the design of the state estimator. It will cause the state estimator to not show sufficient robustness in a high-noise environment, and even lead to system instability. In addition, if the statistical characteristics of errors are not fully analyzed, it is not conducive to reasonably selecting the quantization level, quantization interval and other encoding and decoding parameters, which will cause excessive quantization errors, thus affecting the state estimation accuracy. Especially in application scenarios that require high-precision control, it will lead to a significant decline in the control effect of the system. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for estimating the state of an industrial control system based on an encoding and decoding protocol, which solves the problems in the background art.

[0006] To solve the above technical problems, the present invention adopts the following technical solutions: The present invention provides a method for estimating the state of an industrial control system based on an encoding and decoding protocol, including: Step 1, dual-time scale modeling and encoding and decoding: constructing a discrete-time singular perturbation model, quantifying and encoding the sensor data collected in a specified industrial control process, and analyzing the statistical characteristics of the quantization error.

[0007] Step 2, state estimator design and error analysis: designing a state estimator based on the singular perturbation theory, distinguishing the states of the fast and slow subsystems, and analyzing the exponentially ultimately bounded dynamic error estimated by the state estimator designed by the singular perturbation theory.

[0008] Step 3, LMI solution and stability verification: using the Lyapunov function to analyze the convergence of the quantization error and transforming the stability condition into a linear matrix inequality.

[0009] The beneficial effects of the present invention are as follows: 1. In the embodiment of the present invention, a special form of estimator is designed through a Lyapunov matrix with a singular perturbation parameter and a high degree of freedom, and a set of LMI conditions regarding the singular perturbation parameter are proposed, which ensures the existence of the state feedback controller. When the singular perturbation parameter of the system is within the stable boundary, the state estimation error of the dual-time scale system based on the encoding and decoding protocol converges exponentially ultimately bounded. For the errors generated in the encoding and decoding and communication processes in the industrial control system, a more accurate model is established, so that the data obtained by the state estimator is closer to the actual situation in the industrial control system. Considering the dual-time scale of the industrial control system, designing a state estimator based on the singular perturbation theory can effectively avoid ill-conditioned numerical problems and ensure the boundedness of the system estimation error.

[0010] 2. A method for estimating the state of an industrial control system based on an encoding and decoding protocol provided by the embodiment of the present invention, during the design process of the state estimator, by introducing the condition of exponentially ultimately bounded, it is beneficial to ensure the error convergence and improve the system stability. When designing the state estimator, the document proposes to introduce the condition of exponentially ultimately bounded to ensure that the estimation error can decay to a stable range over time, which is beneficial to ensuring that the system can remain stable during long-term operation and avoiding system instability or failure caused by error accumulation, thus greatly improving the stability and robustness of the system.

[0011] 3. In the process of analyzing the convergence of the quantization error in the embodiment of the present invention, by using the stability analysis method of the Lyapunov function, it is beneficial to the control and evaluation of the system error. Using the Lyapunov function to analyze the error convergence helps the system designer prove the error convergence theoretically. By constructing an appropriate Lyapunov function, it helps to judge whether the quantization error can decay effectively, thereby effectively controlling the propagation of the error.

[0012] 4. When optimizing the gain matrix and the error attenuation rate in the embodiments of the present invention, through the optimization of the gain matrix, it is beneficial to improve the robustness and accuracy of the state estimator. As detailed in the document, by optimizing the gain matrix, the system's response ability to errors is adjusted. With an appropriate gain matrix, the estimation error can be quickly corrected and the influence of quantization error can be effectively reduced. Especially in the face of system noise and uncertainty, the optimization of the gain matrix not only improves the convergence speed of the state estimator but also enhances the system's robustness under perturbations and measurement errors, helping to maintain a high level of accuracy in practical applications and ensuring the reliability of state estimation.

[0013] 5. In the process of system stability analysis in the embodiments of the present invention, by transforming the stability condition into a linear matrix inequality, it is beneficial to accurately solve the optimal gain matrix of the system. Transforming the stability condition into a linear matrix inequality helps to provide an accurate mathematical framework, thereby enabling the designer to accurately solve the optimal gain matrix on the premise of ensuring system stability. This not only improves the accuracy of system design but also helps the designer to achieve optimal performance in the actual control system. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some examples of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0015] Figure 1 It is a schematic diagram of the implementation steps of the present invention.

[0016] Figure 2 It is a state estimation model of an industrial control system based on an encoding and decoding protocol of the present invention.

[0017] Figure 3 It is the norm of the estimation error.

[0018] Figure 4 It is the state x(k) at time k and the estimated state of x(k) at time k

[0019] Figure 5 It is the state x(k + 1) at time k + 1 and the estimated state of x(k + 1) at time k + 1 DETAILED DESCRIPTION OF THE EMBODIMENTS

[0020] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0021] Please refer to Figures 1 to 5 As shown, the present invention provides an industrial control system state estimation method based on an encoding and decoding protocol. The method includes: Step 1, dual-time-scale modeling and encoding and decoding: constructing a discrete-time singular perturbation model, quantifying and encoding the sensor data collected in a specified industrial control process, and analyzing the statistical characteristics of the quantization error.

[0022] It should be noted that Figure 2 the physical object in refers to the actual physical quantities to be monitored and controlled in a specified industrial control system, such as voltage and current, etc. The state estimator refers to the state estimator designed by the singular perturbation theory.

[0023] In a specific embodiment, the process of constructing the discrete-time singular perturbation model is as follows: Divide the specified industrial control system into a fast subsystem and a slow subsystem, and establish singular perturbation models for the fast subsystem and the slow subsystem respectively by the singular perturbation method. Set the singular perturbation index ε, and ε > 0. Then the representation formula of the singular perturbation model corresponding to the fast subsystem is: where k is the time index, representing the number of discrete time steps, f represents the subscript corresponding to the fast subsystem, s represents the subscript corresponding to the slow subsystem, x f (k + 1) represents the state variable corresponding to the fast subsystem at the (k + 1)-th moment, F f (x f (k), x s (k)) represents the function of the fast subsystem's active dynamics, x f (k) represents the state variable corresponding to the fast subsystem at the k-th moment, x s (k) represents the state variable corresponding to the slow subsystem at the k-th moment, g f (x f (k), x s (k)) is the secondary dynamic function of the fast subsystem, w f (k) represents the process noise of the fast subsystem;

[0024] The representation formula of the singular perturbation model corresponding to the slow subsystem is:

[0025] x s (k + 1) = F s (x f (k), x s(k), u(k)) + w s (k), where F s (x f (k), x s (k), u(k)) is expressed as a function of the slow subsystem's active dynamics, w s (k) is expressed as the process noise of the slow subsystem, and then the discrete-time singular perturbation model corresponding to the two-time-scale specified industrial control system is obtained.

[0026] It should be noted that in actual industrial control applications, process noise appears in various control systems, especially in fields such as industrial automation, robot control, aerospace, and transportation systems. For example, suppose there is a temperature control system whose task is to keep the temperature of a certain device or environment within a set range. In an ideal situation, the temperature change of the temperature control system is completely predictable and described by an accurate mathematical model. However, in actual operation, the temperature change is affected not only by control inputs such as the power of the heater and the external environment such as air flow and humidity, but also by many unpredictable factors, and these unpredictable factors constitute the process noise.

[0027] It should be noted that w f (k) is a random variable, usually used to represent unpredictable disturbances in the system, such as environmental noise and measurement errors, etc. ε is set according to the physical characteristics of the industrial control system. For example, the fast subsystem is the voltage or current change in an electrical system, and the slow subsystem is the rotational speed or displacement change in a mechanical system. Through experiments or actual measurement data, the response speeds of the fast and slow subsystems are determined, and thus the magnitude of ε is estimated. If some parts of the system change fast while other parts change slow, then the value of ε is determined according to the actual differences.

[0028] In a specific example, the quantization encoding of the sensor data collected during the specified industrial control process is as follows: During the specified industrial control process, various industrial parameters generated during the specified industrial control process are collected through sensors. If the value range of a certain industrial parameter p is [b, c], then the quantization level L is defined for p, where L represents the number of levels of the quantizer, indicating that the value range of p is divided into L discrete intervals. Then, through the calculation formula: The interval width Δ of the quantization interval corresponding to p is obtained, that is, the minimum change amount represented by each quantization interval corresponding to p. The output after quantization is obtained through the probability quantizer

[0029]

[0030] where is expressed as The calculation result is rounded downwards. According to the number of levels L of the set quantizer, the quantized and mapped value is obtained accordingly. The number of bits for binary coding is log2(L), and thus the quantized and mapped value is obtained. The corresponding coded binary string.

[0031] It should be noted that various industrial parameters include physical quantities such as temperature, pressure, current, and voltage.

[0032] Since random bit flips often occur during the transmission of binary data over the channel, in order to consider this situation in state estimation, this patent will reflect this error in the modeling. Assume that the phenomenon of each bit of the transmitted data being flipped is α i (k), that is

[0033]

[0034] The random variable α i (k) (i = 1, 2,... L) satisfies the following distribution:

[0035] Prob{α i (k) = 1} = δ

[0036] Prob{α i (k) = 0} = 1 - δ

[0037] where δ ∈ [0, 1] is the probability of each binary string bit being flipped. The data obtained after passing through the communication channel can be obtained Obviously, there will be an expected deviation between the data after channel flipping and the original data. Its expectation and variance are

[0038]

[0039] To prevent the deviation between the data and the original data in expectation, artificial compensation is performed on it.

[0040]

[0041] After calculation, the deviation σ(k) between the signal finally reaching the state estimator and the quantized signal, and its expectation and variance are

[0042] E{σ(k)} = 0

[0043]

[0044] The signal finally reaching the state estimator is It can be expressed as

[0045] Since the final average error is 0, no systematic deviation is generated for the specified industrial control system.

[0046] An industrial control system state estimation method based on an encoding and decoding protocol provided by an example of the present invention. During the design process of the state estimator, by introducing the exponentially ultimately bounded condition, it is beneficial to ensure error convergence and improve system stability. When designing the state estimator, the document proposes to introduce the exponentially ultimately bounded condition to ensure that the estimation error can decay to a stable range over time, which is beneficial to ensuring that the system can remain stable during long-term operation and avoiding system instability or failure caused by error accumulation, thereby greatly improving the stability and robustness of the system.

[0047] In the process of analyzing the convergence of quantization errors in an example of the present invention, by using the stability analysis method of the Lyapunov function, it is beneficial to the control and evaluation of system errors. Using the Lyapunov function for error convergence analysis helps system designers theoretically prove the convergence of errors. By constructing an appropriate Lyapunov function, it helps to determine whether the quantization error can effectively decay, thereby effectively controlling the propagation of errors.

[0048] Step 2: State estimator design and error analysis: Design a state estimator based on singular perturbation theory, distinguish the states of the fast and slow subsystems, and analyze the exponentially ultimate boundedness of the estimated dynamic error of the state estimator designed by singular perturbation theory.

[0049] In a specific example, the process of distinguishing the states of the fast and slow subsystems is as follows: The state estimator designed based on singular perturbation theory adopts the following structure: where is the estimated value of the specified industrial control system state at time k + 1, is the output of the state estimator, representing the system state estimated by measuring the input and output information of the system. A is the state transition matrix, indicating the state change of the specified industrial control system from time k to time k + 1. The state transition matrix A projects the current state to the next time and then reflects the dynamic change of the state of the fixed industrial control system. L represents the gain matrix, y(k) represents the observed value collected from the sensor device at time k, C represents the observation matrix, represents the estimated value predicted according to the current state , represents the observation error value between the observed value and the current state estimated value.

[0050] The dynamic response speeds of the fast and slow subsystems are separated through a singular perturbation model, and by setting the gain matrices of the fast and slow subsystems, the state estimations of the fast and slow subsystems are optimized respectively. Thus, through a recurrence formula, a state estimator is designed based on singular perturbation theory to update the states of the fast and slow subsystems cyclically according to the observed values and control inputs, and further estimate the states of the fast and slow subsystems.

[0051] It should be noted that when designing a state estimator based on singular perturbation theory, the state variables of the fast and slow subsystems are distinguished according to the model. In the state estimation of the fast subsystem, due to its fast change speed, in the state transition matrix A, the part of the fast subsystem will be associated with a small singular perturbation index ε, so that the state update of the fast subsystem is frequent and fast. The state of the slow subsystem is estimated at a low frequency due to its slow response, and in the dynamic equation of the slow subsystem, the dynamic change of the part of the A matrix corresponding to the slow subsystem is slow. Therefore, in the design of the state estimator, the adjustment speed of the state estimation of the slow subsystem within each time step is lower than that in the fast subsystem. In the process of recursive update, the states of the fast and slow subsystems are adjusted by the gain matrix L. For the fast subsystem, due to its fast change speed, the gain matrix L plays a more significant role in the estimation process of the fast subsystem, enabling the state of the fast subsystem to be estimated quickly and accurately. For the slow subsystem, the gain matrix L has a small adjustment effect on the estimation of the slow subsystem because the state change of the slow subsystem is slower than that of the fast subsystem, and the state estimator designed by singular perturbation theory has a low adjustment frequency for the error of the slow subsystem.

[0052] In a specific example, the index of the dynamic error estimated by the state estimator designed by analyzing singular perturbation theory is ultimately bounded, and the specific process is as follows: when there exist positive scalars a > 0, 0 ≤ λ < 1, and μ > 0 such that the evaluation formula for the ultimate boundedness of the dynamic error exponent: holds, ||e(k)|| and ||e(0)|| represent the error values corresponding to time k and the initial time, respectively. a is a set constant used to define the error decay rate, indicating that the estimation error will decay exponentially in time. λ is a positive number representing a constant related to the error decay rate, and μ is a parameter related to the initial state of the system or the initial upper bound of the error. μ controls the initial value of the error and is usually used to describe the convergence process of the system from the initial state to the stable state.

[0053] In the example of the present invention, when optimizing the gain matrix and the error attenuation rate, through the optimization of the gain matrix, it is beneficial to improve the robustness and accuracy of the state estimator. As detailed in the document, by optimizing the gain matrix, the response ability of the system to errors is adjusted. With an appropriate gain matrix, the estimation error can be quickly corrected and the influence of quantization error can be effectively reduced. Especially in the face of system noise and uncertainty, the optimization of the gain matrix not only improves the convergence speed of the state estimator, but also enhances the robustness of the system under disturbances and measurement errors, helps to maintain a high level of accuracy in practical applications, and ensures the reliability of state estimation.

[0054] Step 3: LMI solution and stability verification: Use the Lyapunov function to analyze the convergence of the quantization error and transform the stability condition into a linear matrix inequality.

[0055] In a specific example, the process of using the Lyapunov function to analyze the convergence of the quantization error is as follows: Construct the Lyapunov function: where P is a positive definite matrix, e T (k) is the quantization error vector. If the change in the Lyapunov function ΔV(e(k)) at each time step is proven to be negative, then the error of the specified industrial control system will decrease over time and eventually converge, that is, when:

[0056] ΔV(e(k)) = e T (k + 1)P(ε)e(k + 1) - e T (k)P(ε)e(k) < 0 holds, it indicates that the quantization error value corresponding to the specified industrial control system converges.

[0057] In a specific example, the process of transforming the stability condition into a linear matrix inequality is as follows: For the intermediate matrix Φ<0, there exist:

[0058]

[0059] a positive scalar ε0 and symmetric matrices S1, S2, S3 of appropriate dimensions. If S1 ≤ 0, S1 + ε0S2 < 0, holds, then Therefore, the matrix Φ<0 is transformed into:

[0060]

[0061] and where:

[0062] L1 = P 11 K1 + εP 12 K2, Therefore, the parameter estimation of the state estimator designed by the singular perturbation theory is transformed into:

[0063] Since P(ε) is a positive definite matrix, it indicates that: is a non-singular matrix, then When ε→0,

[0064]

[0065] In the process of system stability analysis in the embodiment of the present invention, by transforming the stability condition into a linear matrix inequality, it is beneficial to accurately solve the optimal gain matrix of the system. Transforming the stability condition into a linear matrix inequality helps to provide an accurate mathematical framework, which is beneficial for the designer to accurately solve the optimal gain matrix on the premise of ensuring the system stability. It not only improves the accuracy of system design, but also helps the designer to achieve the optimal performance in the actual control system.

[0066] The embodiment of the present invention designs a set of special estimator forms through a Lyapunov matrix with singular perturbation parameters and a high degree of freedom, and proposes a set of LMI conditions for the singular perturbation parameters, which guarantees the existence of the state feedback controller. When the singular perturbation parameters of the system are within the stability boundary, the state estimation error of the double-time scale system based on the encoding and decoding protocol converges exponentially bounded. For the errors generated in the encoding and decoding and communication processes in the industrial control system, a more accurate model is established, so that the data obtained by the state estimator is closer to the actual situation in the industrial control system. Considering the double-time scale of the industrial control system, designing the state estimator based on the singular perturbation theory can effectively avoid ill-conditioned numerical problems and ensure the boundedness of the system estimation error.

[0067] The above content is only an example and explanation of the concept of the present invention. Those skilled in the art of the present technology can make various modifications or supplements to the described specific examples or use similar methods to replace them. As long as they do not deviate from the concept of the invention or exceed the scope defined by this specification, they should all belong to the protection scope of the present invention.

Claims

1. A method for estimating the state of an industrial control system based on a coding and decoding protocol, characterized in that: include: Step 1: Dual-time scale modeling and encoding: Construct a discrete-time singular perturbation model, quantize and encode the sensor data collected in the specified industrial control process, and analyze the statistical characteristics of the quantization error; Step 2: State estimator design and error analysis: Design a state estimator based on singular perturbation theory, distinguish the states of the fast and slow subsystems, and analyze whether the exponent of the dynamic error estimated by the state estimator designed by the singular perturbation theory is ultimately bounded; Step 3: LMI solution and stability verification: Use the Lyapunov function to analyze the convergence of the quantization error and transform the stability condition into a linear matrix inequality.

2. The method for estimating the state of an industrial control system based on a coding and decoding protocol according to claim 1, characterized in that: The specific process of constructing the discrete time singular perturbation model is as follows: The designated industrial control system is divided into a fast subsystem and a slow subsystem. The singular perturbation model of the fast subsystem and the slow subsystem is established by the singular perturbation method. The singular perturbation index ε is set, and ε>0. The expression formula of the singular perturbation model corresponding to the fast subsystem is: x s (k))+g f (x f (k), x s (k))+w f (k), where k is the time index, indicating the number of discrete time steps, f is the subscript corresponding to the fast subsystem, s is the subscript corresponding to the slow subsystem, and x is the subscript corresponding to the slow subsystem. f (k+1) represents the state variable of the fast subsystem at time k+1, F f (x f (k),x s (k)) is expressed as a function of the main dynamics of the fast subsystem, x f (k) represents the state variable of the fast subsystem at time k, x s (k) represents the state variable of the slow subsystem at time k, g f (x f (k), x s (k)) is the secondary dynamic function of the fast subsystem, w f (k) represents the process noise of the fast subsystem; The expression formula of the singular perturbation model corresponding to the slow subsystem is: s (k+1)=F s (x f (k),x s (k))+w s (k), where F s (x f (k), x s (k)) is expressed as a function of the main dynamics of the slow subsystem, w s (k) is expressed as the process noise of the slow subsystem, and then the discrete-time singular perturbation model corresponding to the dual-time-scale specified industrial control system is obtained.

3. The method for estimating the state of an industrial control system based on a coding and decoding protocol according to claim 2, characterized in that: The specific process of quantizing and encoding the sensor data collected in the specified industrial control process is as follows: In a specified industrial control process, various industrial parameters generated in the specified industrial control process are collected by sensors. If the value range of a certain type of industrial parameter p is [b, c], a quantization level L is defined for p, where L represents the level of the quantizer, indicating that the value range of p is divided into L discrete intervals. The calculation formula is: The width of the quantization interval corresponding to p is obtained, that is, the minimum change represented by each quantization interval corresponding to p. The output obtained after quantization is obtained through the probability quantizer in Expressed as a pair The result of the calculation is rounded down, and the value after quantization mapping is obtained according to the set quantizer level L. The number of bits for binary encoding is log2(L), and the value after quantization mapping is obtained The corresponding encoding secondary string.

4. According to the state estimation method of industrial control system based on codec protocol described in claim 3, since binary data often generates random bit flipping when transmitted in the channel, in order to consider this situation in state estimation, this patent will reflect this error in modeling. Assume that the phenomenon of flipping of each bit of the transmitted data is α i (k), i.e. Random variable α i (k)(i=1,2,...L) satisfies the following distribution: Prob{a i (k)=1}=δ Prob{a i (k)=0}=1-δ Where δ∈[0,1] is the probability of flipping each bit of the binary string. The data obtained after passing through the communication channel can be obtained Obviously, after the channel is flipped, the data will deviate from the original data in the expected way. Its expectation and variance are In order to prevent the data from deviating from the expected value, artificial compensation is performed. After calculation, the deviation σ(k) between the signal that finally reaches the state estimator and the quantized signal and its expectation and variance are obtained as follows: E{σ(k)}=0 The final signal reaching the state estimator is It can be expressed as Since the final error average is 0, no systematic deviation is generated for the specified industrial control system.

5. The method for estimating the state of an industrial control system based on a codec protocol according to claim 4, characterized in that: The specific process of distinguishing the states of the fast and slow subsystems is as follows: The state estimator designed based on singular perturbation theory adopts the following structure: in is the estimated value of the state of the specified industrial control system at time k+1, is the output of the state estimator, which indicates the system state estimated by measuring the input and output information of the system. A is the state transfer matrix, which indicates the state change of the specified industrial control system from time k to time k+1. The state transfer matrix A converts the current state Projection to the next moment It then reflects the dynamic changes of the state of a certain industrial control system. L is represented by the gain matrix, y(k) represents the observation value collected from the sensor device at time k, and C is represented by the observation matrix. Represented as based on the current state The estimated value of the forecast, It is expressed as the observation error between the observed value and the estimated value of the state at the current moment; The dynamic response speed of the fast and slow subsystems is separated by the singular perturbation model, and the state estimation of the fast and slow subsystems is optimized respectively by setting the fast subsystem gain matrix and the slow subsystem gain matrix. Therefore, through the linear matrix inequality formula, the singular perturbation theory is used to design the state estimator to cyclically update the states of the fast and slow subsystems according to the observed values ​​and control inputs, and then the states of the regional fast and slow subsystems.

6. The method for estimating the state of an industrial control system based on a codec protocol according to claim 5, characterized in that: The exponential of the dynamic error estimated by the state estimator designed by the analytical singular perturbation theory is ultimately bounded, and the specific process is as follows: When there exists a positive scalar a>0,0≤λ<1 and μ>0, the exponential of the dynamic error is finally bounded. holds true, ||e(k)|| and ||e(0)|| indicate the error values ​​corresponding to time k and the initial time, a is a constant set to define the error decay rate, indicating that the estimated error will decay exponentially over time, λ is a positive number, representing a constant related to the error decay rate, μ is a parameter related to the initial state of the system or the initial upper bound of the error, μ controls the initial value of the error, and is usually used to describe the convergence process of the system from the initial state to the stable state.

7. The method for estimating the state of an industrial control system based on a codec protocol according to claim 6, characterized in that: The Lyapunov function is used to analyze the convergence of the quantization error. The specific process is as follows: Construct the Lyapunov function: in P is a positive definite matrix, e T (k) is the quantized error vector. If the change of the Lyapunov function ΔV(e(k)) at each time step is proved to be negative, the error of the specified industrial control system will decrease over time and eventually converge, that is, when ΔV(e(k)) = e T (k+1)P(ε)e(k+1)-e T If (k)P(ε)e(k)<0 holds, it means that the corresponding quantization error value of the specified industrial control system converges.

8. The method for estimating the state of an industrial control system based on a codec protocol according to claim 7, characterized in that: The stability condition is transformed into a linear matrix inequality. The specific process is as follows: For the intermediate matrix Φ<0 there exists: Positive scalar ε0 and symmetric matrices S1, S2, S3 of appropriate dimensions, if S1≤0, S1+ε0S2<0, If established, Therefore, the matrix Φ<0 changes to: and in: L1=P 11 K1+εP 12 K2, Therefore, the parameters of the state estimator designed by singular perturbation theory are estimated as follows: Translates to: Since P(ε) is a positive definite matrix, it can be shown that: is a non-singular matrix, then When ε→0,

Images