State Estimation Method for Complex Networks with Switching Topology

By establishing a discrete nonlinear model of time delay based on binary encoding scheme in complex networks, using the Liyapunov stability theorem to solve the gain matrix of the state estimator, which solves the problem that the existing technology cannot consider Markov parameters, sensor nonlinearity and time at the same time. The problem of hysteresis realizes efficient state estimation and network security monitoring of complex network switching topology.

CN115081337BActive Publication Date: 2024-11-05HANGZHOU NORMAL UNIVERSITY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210793633.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-07-05
Publication Date
2024-11-05
Estimated Expiration
2042-07-05

AI Technical Summary

Technical Problem

The existing state estimation method cannot consider Markov parameters, sensor nonlinearity, time delay and random interference at the same time, which affects the state estimation performance, especially in complex networks with switching topology structures, state estimation method under binary encoding scheme Not effectively resolved yet.

Method used

Based on the binary encoding scheme, a discrete nonlinear complex network model with time delay is established, and a state estimation augmentation system is obtained through state estimation error. The state estimator gain matrix is ​​solved using the Liyapunov stability theorem to complete the time delay of the switching topology. State estimation of nonlinear complex network systems.

Benefits of technology

It realizes effective monitoring of the operating status of complex networks and ensures safe and smooth operation of the network, reduces energy consumption during transmission process and the occupation of network bandwidth resources, and is suitable for complex network modeling and state estimation in engineering.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115081337B_ABST
    Figure CN115081337B_ABST
Patent Text Reader

Abstract

The present invention discloses a state estimation method for a complex network with a switching topology. The method of the present invention first establishes a discrete nonlinear complex network model with time lag based on a binary coding scheme, performs state estimation on it, obtains the state estimation error, and then obtains a state estimation augmented system; according to the Lyapunov stability theorem, obtains the state estimator gain matrix; brings the state estimator gain matrix into the state estimator model, and completes the state estimation based on binary coding for the time-lag nonlinear complex network system considering the switching topology. The method of the present invention takes into account complex networks with coupling, time lag and nonlinearity, and can be used to model complex networks in many projects. The method of the present invention can reduce the energy consumption of the transmission process and the occupation of network bandwidth resources, and has important applications in ensuring the safe and smooth operation of the network.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the field of computer technology, in particular to the technical field of state estimation and control methods, and specifically relates to a state estimation method based on binary coding for a time-delay nonlinear complex network with a switching topology structure. Background Art

[0002] Complex networks have attracted the attention of many scholars because they can simulate various large and complex systems such as social networks, power grids, and communication networks. However, the dynamic changes of power grids and communication networks in real life bring great challenges to the analysis of dynamic problems. It is well known that state information plays a vital role in the dynamic analysis of complex systems. However, in most cases, the state information of each node is unknown, so state estimation has become a popular research method, which uses known measurable information to perform state estimation. Due to the rapid development of communication networks, binary coding schemes have become one of the most popular schemes in digital communications. Due to its robustness and ease of implementation in transmission, under binary coding schemes, signals should first be encoded into a set of binary bits and transmitted through binary symmetric channels. Due to the influence of channel noise, this transmission process inevitably produces random bit errors, which makes the decoded measurement signal significantly deviate from the original signal. In recent years, people have conducted preliminary research on binary coding schemes. However, under binary coding, there is no state estimation method for complex networks with switching topology structures. Summary of the invention

[0003] The purpose of the present invention is to provide a state estimation method for a complex network with a switching topology structure to solve the problem that the existing state estimation method cannot simultaneously consider Markov parameters, sensor nonlinearity, time lag, and random interference, thereby affecting the state estimation performance.

[0004] The method of the present invention specifically comprises:

[0005] Step (1) Based on a binary coding scheme, a discrete nonlinear complex network model with time lag is established;

[0006] Step (2) based on a binary coding scheme, performing state estimation on a discrete nonlinear complex network model with time lag to obtain a state estimation error;

[0007] Step (3) obtaining a state estimation augmented system according to the state estimation error;

[0008] Step (4) using the state estimation augmented system, according to Lyapunov stability theorem, obtain the state estimator gain matrix;

[0009] Step (5) brings the state estimator gain matrix into the state estimator model in step (2) to complete the binary-coded state estimation of the time-delay nonlinear complex network system considering the switching topology structure.

[0010] Further, step (1) is specifically:

[0011] The state space form of the discrete nonlinear complex network model is:

[0012]

[0013] in, Indicates the i-th node in the complex network dynamic model The state vector at time, Indicates that the i-th node is The state vector at time, Indicates that the i-th node is The measured output received from the system at the moment; f(·) represents the nonlinear excitation function that satisfies the sector-bounded condition; and represents noise with zero expectation and bounded variance; The time delay of the complex network dynamic system is bounded by a positive scalar τ m and τ M range, Indicates that the jth node is The state vector at time The modality of the moment is a random variable, which is a homogeneous Markov chain in a finite state space range {1,2,…,S}, with the modal number s=1,2,…,S, S is the number of modes, Γ indicates that the internal coupling matrix is ​​a diagonal matrix, and the intermediate function δ(·) is defined as represents the weighted adjacency parameter between the i-th node and the j-th node in the s-th mode in the complex network dynamic model, i, j = 1, 2, ..., N, and i ≠ j, N is the number of nodes, and the weighted adjacency [·] N×N represents an N×N dimensional matrix; A i , C i , D i is a known parameter matrix;

[0014] Measurement output It is first encoded into a finite-length binary string and then transmitted to the estimator for processing through a memoryless binary symmetric channel. The specific process is as follows:

[0015] First measure the output It is encoded into a binary string of length L by the encoder. The amplitude range [-h,h] is divided into 2 L - 1 segment with evenly spaced lengths

[0016] Using random cutoff function For measurement output Preprocessing is performed to obtain the truncated signal Specifically: when J l Represents the value of the set l-th binary character, l = 1, 2, ..., L, and we get: Probability Probability Probability value Truncation error is a random variable, Expectations Variance

[0017] The signal will be cut off Encoded into a binary bit string of length L And the characters pass get.

[0018] Binary bit string Transmitted through a memoryless binary symmetric channel, flipped with crossover probability, the flipped string And the characters is a Bernoulli variable, satisfying The crossover probability α∈[0,1].

[0019] According to the received reversed string The recovered signal after decoding

[0020] Expectations variance Median

[0021] In order to compensate for the distortion, a recovery measurement is performed and the output is measured after recovery.

[0022] Set bit reversal error is a random variable whose expectation variance

[0023] The final measured output of the discrete nonlinear complex network model is

[0024] Furthermore, the state estimator model in step (2) is:

[0025]

[0026] in, Indicates that the i-th node in the state estimator model is Always The estimated value of Indicates that the i-th node is Always The estimated value of Representation and modality The associated state estimator gain matrix to be determined.

[0027] The state estimation error equation is:

[0028] Among them, K i,s Represents the sth mode The associated state estimator gain matrix to be determined.

[0029] The i-th node is The state estimation error at time The i-th node is The state estimation error at time

[0030] Furthermore, the state estimation augmented system in step (3) is expressed as:

[0031]

[0032] in, State estimation error after time augmentation col N {·} represents an N-dimensional column vector, and the state estimator gain matrix K after augmentation is s =diag N {K i,s}, diag N {·} represents an N-dimensional diagonal matrix; the augmented parameter matrix A = col N {A i},D=col N {D i}, C = diag N {C i}, truncation error after augmentation Augmented bit-reversal error Augmented weighted adjacency matrix Intermediate Function Represents the Kronecker product operation.

[0033] Furthermore, step (4) is specifically:

[0034] Depend on Get the matrix P S ,X S , through the formula Solve for the state estimator gain matrix K S ; Among them, P S ,X S is the matrix to be determined, I is the identity matrix, γ and π are both unknown normal constants, T represents the transpose operation, * represents the symmetric terms in the matrix, and the nonlinear coefficient matrix Nonlinear parameter matrix Q represents an unknown symmetric positive definite matrix, U1 and U2 are known constant matrices, and the variation matrix to be determined is Transition probability value Satisfying 0<λ s′s <1 and

[0035] The Lyapunov stability theorem is:

[0036]

[0037] in:

[0038] Function first component Representation and The associated unknown symmetric positive definite matrix;

[0039] Function second component t represents time t;

[0040] Function third component t′ represents time t′.

[0041] In the formula, for Moments and modalities The related Lyapunov function,

[0042] for Moments and modalities Related Lyapunov functions.

[0043] The state estimation method of the present invention takes into account complex networks with coupling, time lag and nonlinearity, and can be used to model complex networks in many projects. In addition, the estimation method of the present invention adopts a binary coding scheme, which can reduce the energy consumption of the transmission process and the occupation of network bandwidth resources, model the system, obtain a nonlinear system model, obtain the overall system model through state augmentation, and finally obtain the parameter design scheme of the estimator. By estimating the state, it is possible to monitor the operating status of the complex network and ensure the safe and smooth operation of the network, which has important applications and thus meets the actual application needs well. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a schematic diagram of the process of the present invention;

[0045] Figure 2 Is the status information on node 1 and estimated trajectory Comparison chart;

[0046] Figure 3 Is the status information on node 1 and estimated trajectory Comparison chart;

[0047] Figure 4 Is the status information on node 2 and estimated trajectory Comparison chart;

[0048] Figure 5 Is the status information on node 2 and estimated trajectory Comparison chart;

[0049] Figure 6 This is the status information on node 3 and estimated trajectory Comparison chart;

[0050] Figure 7 This is the status information on node 3 and estimated trajectory Comparison chart;

[0051] Figure 8 is the estimated error of the first component of each node;

[0052] Fig. 9 is the estimated error of the second component of each node. DETAILED DESCRIPTION

[0053] The present invention is further described below in conjunction with specific embodiments, but the following embodiments are by no means intended to limit the present invention in any way.

[0054] like Figure 1 As shown in FIG. 1 , the state estimation method of a complex network with a switching topology is as follows:

[0055] Step (1) establishes a discrete nonlinear complex network model with time lag based on a binary coding scheme; specifically:

[0056] The state space form of the discrete nonlinear complex network model is:

[0057]

[0058] in, Indicates the i-th node in the complex network dynamic model The state vector at time, Indicates that the i-th node is The state vector at time, Indicates that the i-th node is The measured output received from the system at the moment; f(·) represents the nonlinear excitation function that satisfies the sector-bounded condition; and represents noise with zero expectation and bounded variance; The time delay of the complex network dynamic system is bounded by a positive scalar τ m and τ M range, Indicates that the jth node is The state vector at time The modality of the moment is a random variable, which is a homogeneous Markov chain in a finite state space range {1,2,…,S}, with the modal number s=1,2,…,S, S is the number of modes, Γ indicates that the internal coupling matrix is ​​a diagonal matrix, and the intermediate function δ(·) is defined as represents the weighted adjacency parameter between the i-th node and the j-th node in the s-th mode in the complex network dynamic model, i, j = 1, 2, ..., N, and i ≠ j, N is the number of nodes, and the weighted adjacency [·] N×N represents an N×N dimensional matrix; A i , C i , D i is a known parameter matrix.

[0059] Taking into account the binary coding scheme, the output is measured It is first encoded into a finite-length binary string and then transmitted to the estimator for processing through a memoryless binary symmetric channel. The specific process is as follows:

[0060] First measure the output It is encoded into a binary string of length L by the encoder. The amplitude range [-h,h] is divided into 2 L - 1 segment with evenly spaced lengths

[0061] Using random cutoff function For measurement output Preprocessing is performed to obtain the truncated signal Specifically: when J l Represents the value of the set l-th binary character, l = 1, 2, ..., L, and we get: J l Probability Probability Probability value Truncation error is a random variable, Expectations Variance

[0062] The signal will be cut off Encoded into a binary bit string of length L And the characters pass get.

[0063] Binary bit string Transmitted through a memoryless binary symmetric channel, where each bit may be flipped with a crossover probability due to channel noise, and the flipped string And the characters is a Bernoulli variable, satisfying The crossover probability α∈[0,1].

[0064] According to the received reversed string The recovered signal after decoding

[0065] Expectations variance Median

[0066] With the original measurement output Compared with the actual received signal It is inevitable that a certain degree of distortion will be experienced. In order to compensate for the distortion, a recovery measurement is performed and the output is measured after recovery.

[0067] Set bit reversal error is a random variable whose expectation variance

[0068] The final measured output of the discrete nonlinear complex network model is

[0069] Step (2) performs state estimation on the discrete nonlinear complex network model with time lag based on the binary coding scheme to obtain the state estimation error. The state estimator model is:

[0070]

[0071] in, Indicates that the i-th node in the state estimator model is Always The estimated value of Indicates that the i-th node is Always The estimated value of Representation and modality The associated state estimator gain matrix to be determined.

[0072] The state estimation error equation is:

[0073] Among them, K i,s Represents the sth mode The associated state estimator gain matrix to be determined.

[0074] The i-th node is The state estimation error at time The i-th node is The state estimation error at time

[0075] Step (3) obtains the state estimation augmented system based on the state estimation error. The state estimation augmented system is expressed as:

[0076]

[0077] in, State estimation error after time augmentation col N {·} represents an N-dimensional column vector, and the state estimator gain matrix K after augmentation is s =diag N {K i,s}, diag N{·} represents an N-dimensional diagonal matrix; the augmented parameter matrix A = col N {A i},D=col N {D i}, C = diag N {C i}, truncation error after augmentation Augmented bit-reversal error Augmented weighted adjacency matrix Intermediate Function Represents the Kronecker product operation.

[0078] Step (4) uses the state estimation augmented system to obtain the state estimator gain matrix according to the Lyapunov stability theorem. Specifically:

[0079] Depend on Get the matrix P S ,X S , through the formula Solve for the state estimator gain matrix K S ; Among them, P S ,X S is the matrix to be determined, I is the identity matrix, γ and π are both unknown normal constants, T represents the transpose operation, * represents the symmetric terms in the matrix, and the nonlinear coefficient matrix Nonlinear parameter matrix Q represents an unknown symmetric positive definite matrix, U1 and U2 are known constant matrices, and the variation matrix to be determined is Transition probability value Satisfying 0<λ s′s <1 and

[0080] The Lyapunov stability theorem is:

[0081]

[0082] in:

[0083] Function first component Representation and The associated unknown symmetric positive definite matrix;

[0084] Function second component t represents time t;

[0085] Function third component t′ represents time t′.

[0086] In the formula, for Moments and modalities The related Lyapunov function, for Moments and modalities Related Lyapunov functions.

[0087] Step (5) brings the state estimator gain matrix into the state estimator model in step (2) to complete the binary-coded state estimation of the time-delay nonlinear complex network system considering the switching topology structure.

[0088] The method of the present invention is used to perform simulation verification.

[0089] System parameters:

[0090]

[0091] C3=[0.3 0.3], D1=D2=D3=0.2, A1=[0.060.05], A2=[0.080.05],

[0092] A3=[0.05 0.04].

[0093] The simulation sets the number of nodes in the complex network to be N = 3, the limited state space range is, and the transition probability value is λ 11 =0.4,λ 12 =0.6,λ 21 =0.3,λ 22 = 0.7, internal coupling matrix Γ = 0.1I, time delay Π is the ratio of pi, then τ m =1,τ M = 2. Nonlinear activation function in also and The upper bound of the variance is 0.4, L = 6, π = 0.1, α = 0.01, The initial value of the state is set to:

[0094]

[0095] Solve to get the state estimator gain matrix K S In the following form:

[0096]

[0097] State Estimator Effects:

[0098] Figure 2 Is the status information on node 1 and estimated trajectory Comparison chart; Figure 3 Is the status information on node 1 and estimated trajectory Comparison chart; Figure 4 Is the status information on node 2 and estimated trajectory Comparison chart; Figure 5 Is the status information on node 2 and estimated trajectory Comparison chart; Figure 6 This is the status information on node 3 and estimated trajectory Comparison chart; Figure 7 This is the status information on node 3 and estimated trajectory Comparison chart; Figure 8 is the estimated error of the first component of each node; Fig. 9 is the estimated error of the second component of each node.

[0099] Depend on Figures 2 to 9 It can be seen that for a time-delay nonlinear complex network system with a switching topology structure, the state estimator design method of the present invention can effectively estimate the target state based on the binary coding scheme.

[0100] The present invention relates to a random switching topology structure and a time-delay state estimation method. It should be further explained that steps (2)-(5) in the method of the present invention provide specific methods for implementing the steps, and those skilled in the art can also further improve the steps and methods in details or implementations according to their understanding and the prior art.

Claims

1. A state estimation method for a complex network with a switching topology, characterized in that: Step (1) Based on a binary coding scheme, a discrete nonlinear complex network model with time lag is established; Step (2) based on a binary coding scheme, performing state estimation on a discrete nonlinear complex network model with time lag to obtain a state estimation error; Step (3) obtaining a state estimation augmented system according to the state estimation error; Step (4) using the state estimation augmented system, according to Lyapunov stability theorem, obtain the state estimator gain matrix; Step (5) brings the state estimator gain matrix into the state estimator model in step (2) to complete the state estimation based on binary coding for the time-delay nonlinear complex network system considering the switching topology structure; The state space form of the discrete nonlinear complex network model described in step (1) is: in, Indicates the i-th node in the complex network dynamic model The state vector at time, Indicates that the i-th node is The state vector at time, Indicates that the i-th node is The measured output received from the system at the moment; f(·) represents the nonlinear excitation function that satisfies the sector-bounded condition; and represents noise with zero expectation and bounded variance; The time delay of the complex network dynamic system is bounded by a positive scalar τ m and τ M range, Indicates that the jth node is The state vector at time The modality of the moment is a random variable, which is a homogeneous Markov chain in a finite state space range {1,2,…,S}, with the modal number s = 1,2,…,S, where S is the number of modes, Γ indicates that the inner coupling matrix is ​​a diagonal matrix, and the intermediate function δ(·) is defined as represents the weighted adjacency parameter between the i-th node and the j-th node in the s-th mode in the complex network dynamic model, i, j = 1, 2, ..., N, and i ≠ j, N is the number of nodes, and the weighted adjacency [·] N×N represents an N×N dimensional matrix; A i , C i , D i is a known parameter matrix; Measurement output It is first encoded into a finite-length binary string and then transmitted to the estimator for processing through a memoryless binary symmetric channel. The specific process is as follows: First measure the output It is encoded into a binary string of length L by the encoder. The amplitude range [-h,h] is divided into 2 L - 1 segment, evenly spaced in length Using random cutoff function For measurement output Preprocessing is performed to obtain the truncated signal Specifically: when J l Represents the value of the set l-th binary character, l = 1, 2, ..., L, and we get: Probability Probability Probability value Truncation error is a random variable, Expectations Variance The signal will be cut off Encoded into a binary bit string of length L And the characters pass get; Binary bit string Transmitted through a memoryless binary symmetric channel, flipped with crossover probability, the flipped string And the characters is a Bernoulli variable, satisfying Crossover probability α∈[0,1]; According to the received reversed string The recovered signal after decoding Expectations variance Median In order to compensate for the distortion, a recovery measurement is performed and the output is measured after recovery. Set bit reversal error is a random variable whose expectation variance The final measured output of the discrete nonlinear complex network model is 2. The state estimation method of a complex network with a switching topology structure according to claim 1, characterized in that: The state estimator model in step (2) is: in, Indicates that the i-th node in the state estimator model is Always The estimated value of Indicates that the i-th node is Always The estimated value of Representation and modality The associated desired state estimator gain matrix; The state estimation error equation is: Among them, K i,s Represents the sth mode The associated desired state estimator gain matrix; The i-th node is The state estimation error at time The i-th node is The state estimation error at time 3. The state estimation method of a complex network with a switching topology structure according to claim 2, characterized in that: The state estimation augmented system in step (3) is expressed as: in, State estimation error after time augmentation col N {·} represents an N-dimensional column vector, and the state estimator gain matrix K after augmentation is s =diag N {K i,s }, diag N {·} represents an N-dimensional diagonal matrix; the augmented parameter matrix A = col N {A i },D=col N {D i }, C = diag N {C i }, truncation error after augmentation Augmented bit-reversal error Augmented weighted adjacency matrix Intermediate Function Represents the Kronecker product operation.

4. The state estimation method of a complex network with a switching topology structure as claimed in claim 3, characterized in that: Step (4) is specifically: Depend on Get the matrix P S ,X S , through the formula Solve for the state estimator gain matrix K S ; Among them, P S ,X S is the matrix to be determined, I is the identity matrix, γ and π are both unknown normal constants, T represents the transpose operation, * represents the symmetric terms in the matrix, and the nonlinear coefficient matrix Nonlinear parameter matrix Q represents an unknown symmetric positive definite matrix, U1 and U2 are known constant matrices, and the variation matrix to be determined is Transition probability value Satisfying 0<λ s′s <1 and The Lyapunov stability theorem is: in: Function first component Representation and The associated unknown symmetric positive definite matrix; Function second component t represents time t; Function third component t′ represents the time t′; where, for Moments and modalities The related Lyapunov function, for Moments and modalities Related Lyapunov functions.