Complex network space sampling intermittent containment synchronization control method based on wave equation
By adopting a spatial sampling intermittent restraint synchronization control method based on the wave equation for complex networks, the problems of poor carrying capacity and low control accuracy in traditional methods are solved. Synchronization of heterogeneous networks is achieved, the number of controllers is reduced, and the effects of external disturbances and time delays are suppressed.
Patent Information
- Application Number
- CN202511349224.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-09-22
- Publication Date
- 2026-01-23
AI Technical Summary
In the synchronous control of complex networks, traditional methods have poor capacity to support multiple flexible agents, low control accuracy, and difficulty in handling the effects of external disturbances and time delays. In particular, there is insufficient research on nonlinear wave equations with time delays.
A complex network spatial sampling intermittent restraint synchronization control method based on wave equation is adopted. By constructing a network topology diagram of intelligent agent cluster, designing local controllers and adaptive control laws, and utilizing the spatial sampling information of some nodes, the number of controllers is reduced, thereby achieving synchronization of heterogeneous networks.
It achieves synchronization of heterogeneous networks, reduces the number of controllers, ensures control accuracy, suppresses the effects of external interference and time delay, and improves the effect of network synchronization.
Smart Images

Figure CN121386374A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of control and information technology, and particularly relates to a complex network space sampling intermittent pinning synchronization control method based on wave equation. BACKGROUND
[0002] The distributed parameter system plays an important role in the research of complex networks. Due to the time-space evolution characteristics of the dynamic behavior of nodes, the complex network facing the distributed parameter has attracted much attention. In particular, the complex network model based on wave equation has shown great application potential in many engineering fields, including flexible manipulator, flexible spacecraft, soft tissue machine, and soft robot technology simulating the movement mechanism of the gastropod.
[0003] In the research of dynamic behavior of complex networks, synchronization is an important content. Generally, the synchronization of the drive-response network refers to the norm of the difference between the drive vector and the response vector tending to zero. The drive network is a reference network or a master network. Its dynamic behavior is autonomously generated, which defines the state or behavior mode that the system is expected to reach. In the synchronization problem, it is the "leader" or "template". The response network is a controlled network or a slave network. Its dynamic behavior needs to be adjusted through the designed controller, so that its state is finally synchronized to the state of the drive network. The response network is the "follower". The synchronization research of the drive-response network is widely used in the fields of communication laser equipment, nuclear magnetic resonance, particle pulverizer, etc.
[0004] In the research of synchronization control of complex networks, it is difficult to implement control on each node due to the large number of nodes in the network. Since pinning control plays a "pull one hair and move the whole body" role, it is particularly important in the research of complex network synchronization. In addition, from the perspective of communication time, discontinuous control strategy can save cost and ensure smooth communication of the channel, so it is widely used in the research of complex network synchronization control. As a common discontinuous control strategy, intermittent control, a control strategy combining control time and rest time, has the characteristics of network security risk, low control cost, good convergence, high resource utilization, etc., and has been successfully introduced and applied to complex networks, especially the synchronization problem of space-time complex networks. For example, the synchronization phenomenon of nonlinear systems is studied by using intermittent control; the possibility of synchronization of drive-response chaotic systems with time delay is achieved through intermittent control; the synchronization problem of parabolic directional space-time complex networks is solved through intermittent pinning control; the intermittent pinning strategy for the outer synchronization of hyperbolic coupled systems with time delay has been proposed by using the Lyapunov method of piecewise continuous function.
[0005] The complex network synchronization control strategy based on wave equation often relies on all spatial state information, which is difficult to realize in practice, therefore, the sampling data control is crucial for the practical application of limited dimension controller in distributed parameter system, especially for the flexible structure body with poor bearing capacity, which becomes an important research direction. In recent years, the research on parabolic distributed parameter system has achieved many interesting results, but the research on hyperbolic distributed parameter system is still relatively limited, especially the research on nonlinear wave equation with time delay is still insufficient. Therefore, the strategy combining spatial sampling data control and intermittent control has deep theoretical value and rich practical value in solving the synchronization problem of complex network based on nonlinear wave equation with time delay. SUMMARY
[0006] Therefore, the application provides a complex network spatial sampling intermittent containment synchronization control method based on wave equation, which realizes the synchronization of two heterogeneous networks and reduces the number of controllers while ensuring control accuracy.
[0007] To solve the above technical problems, the application is implemented as follows.
[0008] A complex network spatial sampling intermittent containment synchronization control method based on wave equation, comprising:
[0009] Step one, constructing a network topology graph of the agent cluster, wherein the leader cluster corresponds to the drive network, and the follower cluster corresponds to the response network; according to the characteristics of the time delay wave equation complex network, establishing the state equation and error system model of the drive network and the response network;
[0010] Step two, designing an intermittent containment controller based on spatial sampling information:
[0011] Selecting part of the nodes in the response network as containment control nodes;
[0012] The agent flexible structure has a controllable space, the controllable space is divided into a plurality of sub-regions, the sub-regions are arranged in one dimension in space; the sub-regions correspond to local controllers; the local controllers are constructed by using the comprehensive speed difference of the same sub-regions on the synchronous motion nodes in the drive network and the response network; the gain of the local controller is set as aperiodic intermittent control gain h i (t), to obtain an intermittent containment controller;
[0013] Step three, according to the intermittent containment controller and the error system model, designing an adaptive control law applied to the controller in step two;
[0014] Step four, taking the displacement difference between the synchronous motion nodes in the driving network and the response network as the control target, the control amount of each local controller in the restraint control node is calculated by using the designed intermittent restraint controller and its adaptive control law, and the synchronization of the driving network and the response network is realized.
[0015] Preferably, in the step two, the local controller is constructed as follows:
[0016] In the response network, l nodes are selected as the restraint control nodes, and the set of restraint control nodes is B;
[0017] The controllable space of the node is divided into r sub-regions Ω1~Ω r , wherein the spatial position boundary of the sub-region Ω m is [x m-1 , x m ], m = 1, 2, …, r;
[0018] The corresponding local controller of the sub-region Ω m on the restraint control node i is defined as
[0019]
[0020] , wherein h i (t) is the gain corresponding to the restraint control node i; is the state-based average measure:
[0021]
[0022] or the state-based point measure:
[0023]
[0024] , wherein is the velocity of the controllable space position x on the restraint control node i in the response network, v i (t, x) is the velocity of the controllable space position x on the synchronous motion node corresponding to the restraint control node i in the driving network; t represents the time; the value range of the integration of the controllable space position x is the sub-region Ω m .
[0025] Preferably, the gain of the local controller is set as a non-periodic intermittent control gain h i (t) as follows:
[0026] The discrete time sequence is adopted to divide the control time into time periods, and each time period is divided into a control duration and a remaining time, and the length of the control duration is δ k .
[0027] In the time period [t k , t k+1 , t k + δ k ], the control gain value corresponding to the time t i + δ k is taken as the current control gain value h k (t i + δ i ).
[0028] Preferably, the control gain of the remaining time is 0.
[0029] Preferably, at the initial time of the control time, the control gain is a set value.
[0030] Preferably, the adaptive control law designed in step 4 is:
[0031]
[0032] where q i is a normal number, is the speed of the controllable spatial position x on the response network of the constraint control node i, v i (t, x) is the speed of the controllable spatial position x on the corresponding synchronous motion node of the constraint control node i in the driven network; t represents the time; the value range of the integral of the controllable spatial position x is the entire controllable space Ω; the set parameters ∈, q i > 0.
[0033] Preferably, the driven network and the response network are isomorphic networks, and the method further comprises:
[0034] ① When the controllable space is divided into multiple sub-regions, the upper limit σ of the length of the sub-region satisfies:
[0035]
[0036] where σ is the maximum sub-space length, i.e. σ = max{x m+1 -x m , m = 1,..., r}, λ min (*) represents the minimum eigenvalue of the calculation matrix *, λ max (*) represents the maximum eigenvalue of the calculation matrix *, α2 is the strong damping coefficient matrix in the state equation of the driven network, and H(t) is a gain matrix composed of the non-periodic intermittent control gain h i (t) of the selected l constraint control nodes as diagonal elements;
[0037] ② The external coupling matrix C of the network topology structure satisfies the following conditions:
[0038]
[0039] in, It is to remove The cofactor matrix is obtained by combining the rows and columns corresponding to the l control nodes in the middle, b is the external coupling strength in the state equations of the driving network and the response network, and Γ is the internal connection matrix in the state equations of the driving network and the response network.
[0040]
[0041] In the formula, For spatial length, (*) l This represents the remainder after removing the rows and columns corresponding to the l control nodes in matrix *. Let α1 represent the Hadamard product, κ1 be the diffusion coefficient matrix in the driving network state equation, κ1 be a constant that makes the nonlinear terms in the driving network state equation satisfy the Lipschitz condition, and A be the state matrix of the agent dynamics model in the leader cluster.
[0042] ③Minimum control rate r inf satisfy:
[0043]
[0044] in, δ k It controls the duration, T k The time width of the k-th control time period. λ represents the calculation of the infimum; λ>0 is the equation concerning the convergence rate of the Lyapunov function. The roots, a1 and a2 are constants, a1>a2≥0. It serves as the upper time-delay bound for complex networks of time-delay wave equations;
[0045] Preferably, the driving network and the response network are heterogeneous networks, and the method further includes:
[0046] ① When the controllable space is divided into multiple sub-regions, the upper limit σ of the length of each sub-region satisfies:
[0047]
[0048] Where σ is the length of the largest subspace, i.e., σ = max{x} m+1 -x m ,m=1,...,r},λ min (*) indicates that the minimum eigenvalue of matrix * is calculated, λ max (*) indicates the calculation of the largest eigenvalue of matrix *, and α2 is the strong damping coefficient matrix in the state equation of the driving network. and α1 are diffusion coefficient matrices in the state equations of the response network and the drive network, respectively; H(t) is a non-periodic intermittent control gain h of the selected l pinning control nodes i (t) is a gain matrix composed of diagonal elements, and ε1 is a constant;
[0049] The external coupling matrix C of the network topology satisfies the following condition:
[0050]
[0051] wherein, is a sub-matrix obtained by removing the rows and columns corresponding to the l pinning control nodes in b is the external coupling strength in the state equations of the drive network and the response network, and Γ is the internal connection matrix in the state equations of the drive network and the response network;
[0052]
[0053] wherein, is the spatial length, l denotes a sub-determinant obtained by removing the rows and columns corresponding to the l pinning control nodes in the matrix *; denotes the Hadamard product, α1 is a diffusion coefficient matrix in the state equation of the drive network, κ1 is a constant that makes the non-linear term in the state equation of the drive network satisfy the Lipschitz condition, A is a state matrix of the agent dynamics model in the leader swarm, is a state matrix of the agent dynamics model in the follower swarm;
[0054] The minimum control rate r inf satisfies:
[0055]
[0056] wherein, δ k is the control duration, T k is the time width of the kth control time period, denotes the calculation of the lower bound; λ>0 is the root of the equation about the convergence speed in the Lyapunov function, a1, a2 are constants, a1>a2≥0, is the time delay upper bound of the time-delay dynamic equation complex network;
[0057] Beneficial effects:
[0058] (1) The application solves the problems of poor bearing capacity and low control precision of traditional network control methods for multi-flexible intelligent agents. Starting from the number, time distribution and space distribution of the controller, part of the nodes are selected, and sub-regions are divided on the nodes, an intermittent restraint sampling framework is established, compared with the existing network control method with time and space evolution characteristics, the intermittent restraint control method of the application realizes the synchronization of two heterogeneous networks and reduces the number of controllers, while ensuring the control precision.
[0059] (2) The application limits the upper limit of the length of the sub-region based on Lyapunov stability analysis, and ensures the control precision on the basis of reducing the number of controllers.
[0060] (3) The application solves the problem that the traditional network control method cannot handle the influence of external interference and time delay on the wave equation at the same time, and further affects the network synchronization. By establishing a sufficient condition based on Lyapunov function, the external coupling matrix and the minimum control rate are limited in the condition, the core purpose of realizing the synchronization of two heterogeneous networks and suppressing external interference and time delay is achieved. BRIEF DESCRIPTION OF DRAWINGS
[0061] Figure 1 is the flow chart of the method of the application;
[0062] Figure 2 is the trajectory of the change of the system state with time without the intermittent restraint controller based on space sampling information, in which Y(t, x) is a synchronization error function;
[0063] Figure 3 is the trajectory of the change of the system state with time under the action of the designed intermittent restraint controller based on space sampling information based on average measure, in which Y(t, x) is a synchronization error function;
[0064] Figure 4 is the trajectory of the change of the system state with time under the action of the designed intermittent restraint controller based on space sampling information based on point measure, in which Y(t, x) is a synchronization error function;
[0065] Figure 5 is the trajectory of the change of the gain of the adaptive control law based on average measure;
[0066] Figure 6 is the trajectory of the change of the gain of the adaptive control law based on point measure. DETAILED DESCRIPTION
[0067] The application will be described in detail below in combination with the drawings and examples.
[0068] The application provides a complex network space sampling intermittent containment synchronization control method based on wave equation.
[0069] The agent in the multi-agent cluster is a flexible agent, for example, an invertebrate soft intelligent robot, which has a controllable space on the flexible structure, divides the controllable space into a plurality of sub-regions, the sub-regions are arranged in one dimension in space, each sub-region corresponds to a local controller, and the upper limit of the sub-region division is limited, and only part of the agents in the cluster is controlled during control, so that the control precision is ensured on the basis of reducing the configuration amount of the controller.
[0070] The specific implementation process of the complex network space sampling intermittent containment synchronization control method based on wave equation includes the following steps:
[0071] Step one, construct the network topology graph of the multi-agent cluster, wherein the leader cluster corresponds to the drive network, and the follower cluster corresponds to the response network; according to the time delay wave equation complex network characteristics, the state equation and error system model of the drive network and the response network are established respectively.
[0072] The network topology graph of the multi-agent cluster, the node is the agent, and the edge is the communication connection relationship between the nodes. In the network topology graph, the local network corresponding to the leader cluster is the drive network, and the local network corresponding to the follower cluster is the response network. The drive network and the response network are regarded as time delay wave equation complex networks, and the state equation and the error system model are established.
[0073] There are time delays and external interference factors, and the application hopes to ensure that the two heterogeneous wave equation complex network systems realize the synchronization state through the non-periodic intermittent control of part of the nodes of the network.
[0074] (1) Establish the model of the drive network
[0075] Drive network model:
[0076]
[0077] The boundary condition is:
[0078]
[0079] The initial condition is:
[0080]
[0081] Wherein, x is the spatial position on the agent, Ω is a controllable space on the agent, the controllable space of the application is a long strip space, respectively are controllable space boundary positions, t is a time variable, z i (t, x) is the state of the space position x on the node i in the driving network at t time (i.e., the displacement from the initial point), which is a state variable related to the time variable t and the space variable x.
[0082] n is the degree of freedom of the agent, i∈γ={1, 2, …, N}, N is the total number of nodes of the driving network, the number of nodes of the response network is the same as that of the driving network; denotes the real number field.
[0083] The subscript t represents the first-order partial derivative with respect to t, the subscript tt represents the second-order partial derivative with respect to t, the subscript x represents the first-order partial derivative with respect to x, and the subscript xx represents the second-order partial derivative with respect to x; txx means first-order partial derivative with respect to t and second-order partial derivative with respect to x; for the state z(t, x) at t time, α1 is the diffusion coefficient matrix of the driving network; α2 is the strong damping coefficient matrix of the driving network; α1, α2>0 is a diagonal matrix; is the state matrix of the dynamics model of the agent in the leader swarm; b>0 is the external coupling strength; Γ is the internal connection matrix. The construction of parameters such as b and Γ can refer to the literature: Lei X., Cai S., Jiang S., Liu Z., Adaptive outer synchronization between two complex delayed dynamical networks via aperiodically intermittent pinning control [J], Neurocomputing, 2017, 222: 26-35.
[0084] f: is a nonlinear function that satisfies the Lipschitz condition, f(0, 0)=0, for is the state of the space position x on the node i in the response network at t time, i∈γ, there are constants κ1, κ2>0 such that
[0085] In the initial condition is the initial displacement of the driving network, is the initial velocity of the driving network, n is the degree of freedom of the agent. H 2() denotes the Sobolev space, The second order partial derivative with respect to the variable is a square integrable function, H 1 () denotes the Sobolev space, The first order partial derivative with respect to the variable is a square integrable function.
[0086] τ(t) denotes the time delay, which satisfies is the upper bound of the time delay.
[0087] In the drive network model, the nodes are connected by coupling terms to form a complex network C=(c ij ) N×N is the external coupling matrix, which represents the topology of the network, and the element c ij >0 represents the connection between node j and node i; based on the conservation of network flow,
[0088] (2) Establish a model of the response network
[0089] In the response network model, a tilde is added to z to distinguish it from z in the drive network.
[0090] Response network model:
[0091]
[0092] The boundary condition is
[0093]
[0094] The initial condition is
[0095]
[0096] where, is the state of node i in the response network at time t; is the diffusion coefficient matrix of the response network; α2 is the strong damping coefficient matrix of the response network, which is the same as the drive network; is also a diagonal matrix; is the state matrix of the dynamic model of the agent in the follower swarm. u i (t) is the controller of node i.
[0097] In the initial condition, is the initial displacement of the response network, is the initial velocity of the response network; Here for the sake of clarity, the (s, x) is dropped.
[0098] In the above response network model, the nodes are connected by coupling terms , forming a complex network C = (c ij ) N×N is the external coupling matrix, representing the topology of the network, and the element c ij > 0 represents the connection of node j with node i,
[0099] (3) Establish the error system model of the driving network and the response network
[0100] Error system:
[0101] Let:
[0102]
[0103] Then:
[0104]
[0105] where,
[0106] is the difference of the nonlinear term, and Θ(·) is the parameter difference term.
[0107] where v i (t, x) is the velocity of the spatial position x on node i in the driving network; is the velocity of the spatial position x on node i in the response network; e i (t, x) is the position difference between the driving network and the response network; is the velocity difference between the driving network and the response network, I n is the n-order unit matrix.
[0108] Step two, design an intermittent restraint controller based on spatial sampling information:
[0109] Select l network nodes in the response network as restraint control nodes to form set B.
[0110] The controllable space of the network node corresponding to the intelligent agent flexible structure has controllable space, which is divided into multiple sub-regions, and the sub-regions are arranged in one dimension in space; each sub-region corresponds to a local controller; the local controller is constructed by using the comprehensive velocity difference of the same sub-region on the synchronous motion nodes in the driving network and the response network; the gain of the local controller is set to the non-periodic intermittent control gain h i (t).
[0111] ① The controllable space Ω of the network node is divided into r sub-regions Ω1 ~ Ω r .
[0112] The controllable space of the present application is a long strip space, ignoring its width, the length range is l 1 and are the upper and lower limits of the controllable space, respectively. The space interval is divided into r sub-intervals Let σ be the maximum length of the sub-interval, Ω m = [x m-1 , x m ).
[0113] ②The sub-area Ω m on the control node i is restrained, and the corresponding local controller is defined as
[0114]
[0115] Where, h i (t) is the gain corresponding to the control node i, which is the non-periodic intermittent control gain. The average measure based on state or the point measure based on state is adopted.
[0116] Average measure scheme: the integrated speed difference of the same sub-area on the synchronous motion node in the driving network and the response network is used to construct the local controller, which is the average measure based on state:
[0117]
[0118] Point measure scheme:
[0119]
[0120] Where, is the speed of the controllable space position x on the control node i in the response network, v i (t, x) is the speed of the controllable space position x on the corresponding synchronous motion node of the control node i in the driving network; t represents the time; the value range of the integral of the controllable space position x is the sub-area Ω m .
[0121] In the above local controller , h i (t) adopts non-periodic intermittent control gain, and the specific setting method is as follows:
[0122] Discrete time sequence The control time is divided into time periods, each of which is divided into a control duration and a remaining time, the control duration has a length of δ k ;
[0123] At the initial moment of the control time, the control gain is a set value;
[0124] At the control duration of the time period [t k ,t k+1 ), the control gain value corresponding to the moment t k + δ k is used as the current control gain value h i (t k + δ k ).
[0125] The control gain of the remaining time is 0.
[0126] The formula of the aperiodic intermittent control gain h i (t) is:
[0127]
[0128] Wherein h i (0) is a normal number, is a discrete time sequence, 0=t1<t2<…<t k <…,δ k is the control duration.
[0129] Step three, according to the intermittent restraint controller and the error system model, the comprehensive speed difference of the controllable space of the synchronous motion node in the driving network and the response network is used to design an adaptive control law Applied to the controller in step two.
[0130] The adaptive control law designed in this step is:
[0131]
[0132] Wherein, q i is a normal number, is the speed of the controllable space position x on the restraint control node i in the response network, v i (t,x) is the speed of the controllable space position x on the corresponding synchronous motion node of the restraint control node i in the driving network; t represents the moment; the value range of the integral of the controllable space position x is the entire controllable space Ω; the set parameters ∈, q i >0.
[0133] Step four: the driving network and the response network are homogenous networks, and a determination condition for realizing synchronization between the driving network and the response network is determined based on a Lyapunov function.
[0134] When the driving network and the response network are homogenous networks, the parameter conditions to be met include:
[0135] ① When the controllable space is divided into multiple sub-regions, the upper limit σ of the length of the sub-region satisfies:
[0136]
[0137] wherein σ is the maximum sub-space length, i.e. σ = max {x m+1 -x m , m = 1,..., r}, λ min (*) represents the minimum eigenvalue of the calculation matrix *, λ max (*) represents the maximum eigenvalue of the calculation matrix *, α2 is a strong damping coefficient in the state equation of the driving network, and H is a gain matrix composed of l non-periodic intermittent control gains h i (t) of the selected control nodes;
[0138] ② The external coupling matrix C of the network topology structure satisfies the following condition:
[0139]
[0140] wherein is a residual sub-matrix obtained by removing the rows and columns where the l nodes corresponding to the set B in are located, b is a coupling strength in the state equation of the driving network and the response network, and Γ is an internal connection matrix in the state equation of the driving network and the response network;
[0141]
[0142] wherein is a space length, and l represents a residual sub-matrix obtained by selecting the rows and columns of the matrix * where the l nodes corresponding to the set B are located; represents a Hadamard product, α1 is a diffusion coefficient in the state equation of the driving network; κ1 is a constant for making the non-linear term in the state equation of the driving network satisfy a Lyapunov condition; and A is a state matrix of the dynamic model of the intelligent agent in the leader swarm.
[0143] ③ The minimum control rate r inf satisfies:
[0144]
[0145] wherein, δ k is the control duration, T k = t k+1 -t k is the time width of the kth time interval, denotes the calculation of the infimum; λ > 0 is the root of the equation about the convergence speed of Lyapunov function , a1, a2 are constants, a1 > a2 ≥ 0, is the time delay upper bound of the time delay fluctuation equation complex network;
[0146] The following explains the specific process of obtaining the determination condition for realizing the synchronization of the drive network and the response network:
[0147] The definition of exponential synchronization is given as follows: if there exist constants ε > 0 and v ≥ 1 such that
[0148]
[0149] then the drive network and the response network are exponentially synchronized.
[0150] In the formula, sup represents the supremum;
[0151] Lyapunov stability theory is: Lyapunov function V(t) satisfies
[0152]
[0153] wherein a1, a2, a3, a4 are constants, k ∈ Z + If there exist constants such that the inequality is established
[0154] a1 > a2 ≥ 0, δ = a1 + a3 > 0
[0155]
[0156] then
[0157]
[0158] wherein t ≥ 0, λ > 0 is the root of the equation . μ is a constant greater than or equal to 0.
[0159] The determination condition for realizing the synchronization of the isomorphic network is:
[0160]
[0161] wherein,
[0162]
[0163] Proof process:
[0164] Lyapunov function is designed as:
[0165]
[0166] wherein Φ(t) is
[0167]
[0168] wherein q i , Derivation of Lyapunov function with respect to time t, using some mathematical inequalities, Schur's complement and the designed controller, obtains:
[0169] t k-1 ≤t≤t k-1 +δ k ,k∈Z + ,
[0170]
[0171] Further, according to Lyapunov stability theory, two networks can achieve exponential synchronization.
[0172] Step five: the drive network and the response network are heterogeneous networks, based on Lyapunov function, determine the judgment condition for realizing synchronization of the drive network and the response network.
[0173] When the drive network and the response network are heterogeneous networks, the parameter conditions that need to be met include:
[0174] ① When the controllable space is divided into multiple sub-regions, the upper limit σ of the length of the sub-region satisfies:
[0175]
[0176] wherein, σ is the minimum sub-space length, λ min (*) represents the minimum eigenvalue of the calculation matrix *, λ max (*) represents the maximum eigenvalue of the calculation matrix *, and α2 is the strong damping coefficient in the state equation of the drive network; and α1 are the diffusion coefficient matrices in the state equations of the response network and the drive network respectively; ε1 is a normal number.
[0177] ② The external coupling matrix C of the network topology structure satisfies the following conditions:
[0178]
[0179] wherein, is a sub-matrix obtained by removing rows and columns corresponding to nodes in set B in matrix b is the coupling strength in the state equation of the drive network and the response network, and Γ is the internal connection matrix in the state equation of the drive network and the response network;
[0180]
[0181] wherein, L is the spatial length, l denotes a sub-determinant obtained by removing rows and columns corresponding to l nodes in set B in matrix denotes the Hadamard product, a1 is a diffusion coefficient in the state equation of the drive network; κ1 is a constant that makes the non-linear term in the state equation of the drive network satisfy the Lipschitz condition; A is a determining matrix, describing the internal dynamic behavior of a node itself;
[0182] ③ minimum control rate r inf satisfies:
[0183]
[0184] wherein, δ k is the control duration, T k = t k+1 -t k is the time width of the kth control time period, denotes calculating the lower bound; λ>0 is the root of equation a1, a2 are constants, a1>a2≥0, is the time delay upper bound of the time-delayed dynamic equation complex network;
[0185] The following explains the specific process of obtaining the determination condition for realizing synchronization of the heterogeneous drive network and the response network:
[0186] The definition of the heterogeneous network exponential synchronization is the same as that in step four. The determination condition for realizing the heterogeneous network synchronization is:
[0187]
[0188] wherein,
[0189]
[0190] H * is a matrix obtained by removing rows and columns corresponding to nodes in set B in matrix Diagonal matrix composed of B, i∈B. The spatial sampling interval satisfies:
[0191]
[0192] Proof process:
[0193] Lyapunov function is designed as:
[0194]
[0195] In the formula,
[0196] Similar to the last proof process, it can be obtained
[0197]
[0198] Further, the two networks can realize exponential synchronization by using Lyapunov stability theory.
[0199] Step six, taking the displacement difference of the synchronous motion node in the driving network and the response network as the control target, using the designed intermittent restraint controller and the adaptive control law, the control amount of each local controller in the restraint control node is solved, and the synchronization of the driving network and the response network is realized.
[0200] At this point, the process is completed.
[0201] Example one:
[0202] The method described in the application is simulated:
[0203] Considering the properties of describing magnetic materials, a one-dimensional damped sine Gordon equation as node dynamics for studying magnetic multi-agent systems is used, which has wide application in the field of medicine, environmental monitoring and intelligent materials:
[0204]
[0205] The boundary condition is zero, and the initial condition is
[0206] Wherein
[0207] z(t,x)=(z1,z2,z3) T (t,x), α1=0.3I3, α2=0.15I3, A=2I3,
[0208] τ=0.2|cos(t)|, M1=M2=0.3I3, k1=k2=0.18,
[0209] f(zi (t,x),z i (t-τ(t),x))=M1sin(z(t,x))+M2sin(z(t-τ(t),x)).
[0210] Based on the scale-free distribution of Barabsi-Albert network model, consider a network with seven nodes, network parameters satisfy
[0211]
[0212] b=20,Γ=0.5I3.
[0213] The effect of the constructed spatial sampling intermittent containment controller:
[0214] Figure 2 is the state trajectory of the error system without the effect of the controller i (t),i=1,2, thus it can be known that under the influence of time delay and external disturbance, the drive network and the response network cannot realize synchronization. The first three nodes are selected to be controlled, the control rate is 0.8, a=0.001, h1=14, h2=12, h1=13, Figure 3 and Figure 4 are respectively the synchronization error function trajectory Y(t,x) of the two networks with time delay and external disturbance under the drive of the intermittent containment controller based on spatial sampling information based on average measure and point measure, according to Figures 2-4 it can be obtained that even if the network is influenced by time delay and external disturbance, the two networks can still realize synchronization under the designed intermittent containment controller based on spatial sampling information. Figure 5 and Figure 6 are respectively the change trajectory of the gain of the adaptive control law based on average measure and point measure, obviously the gain gradually converges to a constant.
[0215] The above specific embodiments only describe the design principles of the application, the shapes and names of the components in the description can be different and are not limited. Therefore, the person skilled in the art of the application can modify or equivalently replace the technical solutions recorded in the foregoing embodiments; and these modifications and replacements do not deviate from the purpose and technical solutions of the application, and should all belong to the protection scope of the application.
Claims
1. A complex network space sampling intermittent containment synchronization control method based on wave equation, characterized in that, The application relates to a method for realizing synchronization of a drive network and a response network. Step one, constructing a network topology graph of an agent cluster, wherein a leader cluster corresponds to the drive network and a follower cluster corresponds to the response network; According to the complex network characteristics of the time-delay wave equation, state equations and error system models of the drive network and the response network are established; Step two, designing an intermittent containment controller based on spatial sampling information: Selecting part nodes in the response network as containment control nodes; The intelligent body flexible structure has a controllable space, the controllable space is divided into multiple sub-regions, the sub-regions are arranged in one dimension in the space; the sub-regions correspond to local controllers; the local controllers are constructed by using the comprehensive speed difference of the same sub-regions on the synchronous motion nodes in the driving network and the response network; the gain of the local controller is set as a non-periodic intermittent control gain h i (t), an intermittent restraint controller is obtained; Step three, according to the intermittent restraint controller and error system model, using the drive network and response network in the synchronous motion node controllable space of the comprehensive speed difference, design adaptive control law Applied to the controller in step two; Step four, taking the displacement difference between the synchronous motion nodes in the drive network and the response network as a control target, using the designed intermittent containment controller and an adaptive control law, solving the control amount of each local controller in the containment control nodes, and realizing the synchronization of the drive network and the response network.
2. The method of claim 1, wherein, In step two, the local controller is constructed in the following manner: Selecting l nodes in the response network as containment control nodes, and the set of the containment control nodes is B; The controllable space of the node is divided into r sub-regions Ω1~Ω r , wherein the spatial position boundary of the sub-region Ω m is [x m-1 , x m ), m=1, 2,..., r; Subregion Ω on the holding control node i m The corresponding local controller is defined as where h i (t) is the gain corresponding to the containment control node i; is the state-based average measure: Or a state-based point measure: wherein wherein v(t, x) is the velocity of the controllable spatial position x on the synchronization motion node corresponding to the containment control node i in the network in response to the velocity of the controllable spatial position x on the containment control node i in the network i v(t, x) is the velocity of the controllable spatial position x on the synchronization motion node corresponding to the containment control node i in the network in response to the velocity of the controllable spatial position x on the containment control node i in the network m v(t, x) is the velocity of the controllable spatial position x on the synchronization motion node corresponding to the containment control node i in the network in response to the velocity of the controllable spatial position x on the containment control node i in the network 3. The method of claim 1, wherein, The gain of the local controller is set to the aperiodic on-off control gain h i (t) is: Using discrete time series dividing the control time into time segments, each time segment being divided into a control duration and a remaining time, the control duration having a length of δ k ; During the time period [t] k ,t k+1 The control duration is t. k +δ k The control gain value at time t is used as the current control gain value h. i (t k +δ k ).
4. The method of claim 3, wherein, The control gain of the remaining time is 0.
5. The method of claim 3, wherein, At the initial moment of the control time, the control gain is a set value.
6. The method of claim 3, wherein, Step 4 Designed Adaptive Control Law is: where q i is a positive constant, is the velocity of the controllable spatial position x on the containment control node i in the response network, i (t, x) is the velocity of the controllable spatial position x on the corresponding synchronous motion node of the containment control node i in the driving network; t represents the time; the value range of the integration of the controllable spatial position x is the entire controllable space Ω; the parameters ∈, q i > 0.
7. The method of claim 1, wherein, The drive network and the response network are homogeneous networks, and the method further comprises: When the controllable space is divided into multiple sub-regions, the upper limit sigma of the length of the sub-region satisfies: where σ is the maximum subspace length, i.e. σ = max{x m+1 - x m , m = 1,..., r}, λ min (*) denotes the smallest eigenvalue of the matrix *, λ max (*) denotes the largest eigenvalue of the matrix *, a2is a strong damping coefficient matrix in the state equation of the drive network, H(t) is a gain matrix consisting of the non-periodic intermittent control gains h i (t) of the selected l number of draw control nodes as diagonal elements; The external coupling matrix C of the network topology structure satisfies the following condition: wherein, is a submatrix obtained by removing the rows and columns corresponding to the l holding control nodes in b is the external coupling strength in the state equation of the drive network and the response network, and Γ is the internal connection matrix in the state equation of the drive network and the response network; wherein is the spatial length, l denotes the minor of the matrix * without the rows and columns corresponding to the l binding control nodes; denotes the Hadamard product, α1 is a diffusion coefficient matrix in the state equation of the drive network; κ1 is a constant that makes the nonlinear term in the state equation of the drive network satisfy the Lipschitz condition; A is the state matrix of the agent dynamics model in the leader swarm. ③ Minimum control rate r inf satisfies: wherein δ k is the control duration, T k is the time width of the kth control period, denotes the computation of the lower bound; λ > 0 is the root of the equation , a1, a2 are constants, a1 > a2 ≥ 0, is the time delay upper bound of the time-delay oscillation equation complex network; 8. The method of claim 1, wherein, The drive network and the response network are heterogeneous networks, and the method further comprises: When the controllable space is divided into multiple sub-regions, the upper limit sigma of the length of the sub-region satisfies: where σ is the maximum subspace length, i.e., σ = max{x m+1 - x m , m = 1,..., r}, λ min (*) denotes the smallest eigenvalue of the matrix *, λ max (*) denotes the largest eigenvalue of the matrix *, and α2is the strong damping coefficient matrix in the state equation of the driving network. and α1are the diffusion coefficient matrices in the state equations of the response network and the driving network, respectively; H(t) is a gain matrix composed of the non-periodic intermittent control gains h i (t) of the selected l number of draw control nodes as diagonal elements, and ε1is a constant. The external coupling matrix C of the network topology structure satisfies the following condition: wherein, is a submatrix obtained by removing the rows and columns corresponding to the l holding control nodes in b is the external coupling strength in the state equation of the drive network and the response network, and Γ is the internal connection matrix in the state equation of the drive network and the response network; In the formula, For the spatial length, (*) l Denotes the minor of the matrix * without the rows and columns corresponding to the l constraint control nodes; Denotes the Hadamard product, α1 is a diffusion coefficient matrix in the state equation of the driving network; κ1 is a constant that makes the nonlinear term in the state equation of the driving network satisfy the Lipschitz condition; A is a state matrix of the agent dynamics model in the leader swarm, A state matrix of the agent dynamics model in the follower swarm Minimum control rate r inf satisfies: wherein δ k is the control duration, T k is the time width of the kth control period, denotes the computation of the lower bound; λ > 0 is the root of the equation , a1, a2 are constants, a1 > a2 ≥ 0, is the time delay upper bound of the time-delay oscillation equation complex network;
