A robust synchronization control method based on improved extended high-gain observer

By improving the expanded high-gain observer and sliding mode control technology, a robust synchronization control protocol is designed, which solves the synchronization control problem caused by model uncertainty and interference in multi-agent systems, and realizes the robustness and synchronization of the system .

CN115016292BActive Publication Date: 2024-12-03UNIV OF ELECTRONICS SCI & TECH OF CHINA
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210830118.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-07-15
Publication Date
2024-12-03
Estimated Expiration
2042-07-15

AI Technical Summary

Technical Problem

Multi-agent systems are difficult to achieve robust synchronous control when facing model uncertainty and unknown external interference, especially because sensor limitations cannot feedback higher-order information items, such as speed information.

Method used

Using an improved expansion high-gain observer combined with continuous sliding mode control technology, a robust synchronization control protocol is designed through singular perturbation theory to estimate unmeasurable speed and interference, and a control input is constructed through the sliding mode surface to achieve system synchronization.

Benefits of technology

It improves the control robustness of multi-agent systems, can realize the synchronization of individual states in the presence of interference and unpredictable states, expands the application scenarios of the algorithm, and ensures that interference and speed estimation errors tend to be bounded.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115016292B_ABST
    Figure CN115016292B_ABST
Patent Text Reader

Abstract

The present invention discloses a robust synchronization control method based on an improved extended high-gain observer. First, a second-order dynamic model of individuals in a multi-agent system is established. For the model, an improved extended high-gain observer is designed, and then a robust synchronization control protocol is designed. A fast and slow dynamic model based on singular perturbations is constructed, and a robust synchronization controller based on continuous sliding mode theory is designed, so that individuals in the multi-agent system can finally achieve state synchronization. The method of the present invention focuses on the adverse effects of interference and unmeasurable speed on the multi-agent system, laying a foundation for technicians to simultaneously solve the adverse effects of interference and unmeasurable speed in actual operation, ensuring that the interference estimation error and the state estimation error tend to a bounded range at the same time, and at the same time, it can analyze the relationship between the convergence of the system synchronization error and the control parameters based on the fast and slow state equations, promote parameter adjustment, improve the effectiveness and robustness of the synchronization control scheme, and expand the application scenario of the algorithm.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of robust synchronization control of multi-agent systems, and in particular relates to a robust synchronization control method based on an improved extended high-gain observer. Background Art

[0002] Group collaboration is widely present in biological populations in nature, for example: ant colonies collaborate to carry food, fish schools swim together to avoid sharks, geese fly in formation to reduce air resistance, and wolves collaborate to hunt prey. Such group phenomena have attracted widespread attention from researchers due to their characteristics of collaboration, intelligence emergence, spontaneity, and dispersion. With the deepening of scientific research, people have a deeper and more intuitive scientific understanding of the internal mechanisms of such biological behaviors. A large number of actual observations and data analyses have shown that although the capabilities of individuals within a group are limited, through global / local information interaction and resource sharing, the group can complete more complex behaviors, such as maintaining formation, avoiding natural enemies, coordinated attacks, and finding food, reflecting high-order intelligence. In these studies, the synchronous control of the group is the basis.

[0003] Synchronous control refers to how to design a distributed interaction protocol so that the individuals in the final system reach the same (motion) state. However, in actual systems, interference, including model uncertainty and unknown external interference, is always widespread and has a serious impact on the control performance of the system. Multi-agent systems will inevitably encounter the above problems. Generally speaking, one way to solve such problems is to design a robust synchronous controller.

[0004] In view of the above problems, by studying the robust synchronization control problem of multi-agent systems, it is found that many existing studies have attempted to solve the problem of robust synchronization control. For example, (1) feedback control, that is, improving the performance of the system by suppressing interference, among which robust adaptive control and variable structure control are typical; (2) feedforward control, that is, estimating the amount of interference through detection, and compensating for the interference through control input in the form of generating feedforward terms. Typical methods include disturbance observers, anti-disturbance controllers, etc. However, in previous studies, the unmeasurability of the system's high-order state was not taken into account. In actual systems, due to the lack of sensors or insufficient sensor accuracy, high-order information items are often unable to be fed back. For example, for a second-order system, speed information cannot be fed back. Summary of the invention

[0005] To solve the above technical problems, the present invention is based on an improved expanded high-gain observer, combined with continuous sliding mode control technology, and through singular perturbation theory, a method for synchronous control of a multi-agent system is proposed, which improves the robustness of system control, expands the application scenarios of the algorithm, and can simultaneously solve the adverse effects of unpredictable states and interference on the system.

[0006] The technical solution adopted by the present invention is: a robust synchronization control method based on an improved extended high-gain observer, and the specific steps are as follows:

[0007] S1. Establish the second-order dynamics model of individuals in the multi-agent system in the field of drones and unmanned vehicles, including two channels: position and velocity;

[0008] S2. For the second-order dynamic model, an improved extended high-gain observer is designed to estimate the unmeasurable speed and unknown disturbance through a parameter-adjustable dynamic system;

[0009] S3, define the synchronization error and design the sliding surface, compensate the disturbance estimation value obtained in step S2 into the controller and replace the unavailable state in the control input with the estimated unmeasurable state, and then design a robust synchronization control protocol;

[0010] S4, bringing the control protocol design established in step S3 into the original dynamic model to obtain a closed-loop error system, and then defining auxiliary variables, and then based on the singular perturbation technology, converting the closed-loop error system into a dynamic equation of fast and slow dynamic coupling to determine the upper bound of the synchronization error;

[0011] S5. Apply the control protocol to the second-order dynamics model of the multi-agent system so that the individuals in the multi-agent system can eventually achieve state synchronization.

[0012] Furthermore, the step S1 includes the following sub-steps:

[0013] S11. The multi-agent system has n individuals, which are connected through an undirected network G. The dynamic model of n second-order individuals is as follows:

[0014]

[0015] in, Represent the position, velocity and control input of the ith individual, represents the real number space, Respectively represent the position derivative and velocity derivative of the ith individual, t represents time, and f i (r i (t),v i (t), t) represents the interference in the control input channel. Due to the limitation of sensors, the speed information of the individual itself cannot be fed back for designing the control protocol. At the same time, an individual in the multi-agent system can only interact with its neighbors (i.e., there is a connection) to complete the information exchange.

[0016] S12. To achieve synchronous control of multi-agent systems, the following definitions are made:

[0017]

[0018] That is, the states of any two individuals in the multi-agent system converge and eventually tend to be consistent. When the conditions shown in (3) are met, the multi-agent system is said to have achieved synchronous control, where r j (t) represents any individual among the n individuals in the multi-agent system except the i-th individual.

[0019] Furthermore, the step S2 includes the following sub-steps:

[0020] S21. First, in order to ensure the effective estimation of interference, it is necessary to make assumptions about the upper bounds of interference and its derivatives.

[0021]

[0022] in, Represents the upper bound of the disturbance and the disturbance derivative, which is a positive number. To control the interference f in the input channel i (r i ,v i ,t), and define the estimated values ​​of individual position, velocity and disturbance in the multi-agent system as The corresponding estimation error is:

[0023]

[0024] in, is the corresponding estimation error, and then for individual i in the multi-agent system, an improved extended high-gain observer is designed in the following form:

[0025]

[0026] in, Represent the corresponding variable estimates The derivative of , where the parameters α1, α2, α3 are selected so that the characteristic polynomial s 3 +α1s 2 +α2s+a3=0 all have negative real parts, where s represents the time domain. In addition, the dynamically adjustable parameter ε is the output state of the following dynamic system:

[0027]

[0028] Among them, ω represents the system state of the dynamic system, is the derivative of ω, ω(0)=ω0,ω0>ε f , ω0 is a constant, ω(0) represents the initial value of the variable ω, ε f is the preset value of the dynamic parameter, and εf is a relatively small quantity. In addition, the parameters c1>0, c2>0, 0<λ1<1 and λ2>1, c1 and c2 represent the proportional gain values, and λ1 and λ2 represent the exponential gain values. Furthermore, based on the finite time theory, it can be known that by using the dynamic system shown in (5), the dynamically adjustable parameter ε can reach ε in a fixed time. f , and the corresponding adjustment time is:

[0029]

[0030] Among them, T Δ A variable representing the time it takes for the system to converge.

[0031] S22. When the extended high-gain observer shown in (4) is used, the simultaneous estimation of the unmeasurable state and the unknown interference can be achieved, and the corresponding estimation performance can be adjusted through the dynamic system (5).

[0032] S23. Introduce the coordinate transformation relationship shown below:

[0033]

[0034] Among them, η 1i , η 2i , η 3i Represent the corresponding estimated error variables The conversion amount, combined with formula (4), obtains the following fast-changing time dynamic system:

[0035]

[0036] Among them, η i =[η 1i ,η 2i ,η 3i ] T , representing three variables η 1i ,η 2i ,η 3i The column vector formed, T represents the matrix transpose, To control the interference f in the input channel i The derivative of η represents the system matrix, B η represents the input matrix. For the dynamic system (6), if the system matrix A η is Hurwitz's, and Lyapunov's equation If it holds true, P and Q are matrices of corresponding dimensions, L η is a positive constant. Then we know that the dynamically adjustable parameter ε satisfies:

[0037]

[0038] In the formula, λ max ,λ min are the maximum and minimum eigenvalues ​​of the corresponding matrix respectively, then we can further obtain that the speed estimation error and interference estimation error satisfy:

[0039]

[0040] in, They represent the upper bound functions of the estimated errors of individual speed and disturbance in the multi-agent system, respectively, and there is Furthermore, it can be seen that the improved extended high-gain observer can effectively achieve the effective simultaneous estimation of unmeasurable states and unknown interferences.

[0041] Furthermore, the step S3 includes the following sub-steps:

[0042] S31, synchronization error is defined as:

[0043]

[0044] Among them, a ij Represents the adjacency matrix in a multi-agent network topology.

[0045] S32. To achieve the convergence of the synchronization error in S31, the following coordinate transformation relationship is designed:

[0046]

[0047] Among them, p 1i 、p 2i Represent the position error variable and speed variable respectively, and we can get is the speed variable p 2i The derivative of .

[0048] S33. Then, based on the variables defined in (8), the sliding surface of the following form can be constructed:

[0049]

[0050] Among them, s i represents the sliding surface variable, k1 and k2 represent the proportional gain parameters, both of which are positive numbers. Then, the derivative of (9) is:

[0051]

[0052] S34. Based on the classical sliding mode technology, the following sliding mode surface approach law can be constructed:

[0053]

[0054] Among them, ζ, ks represents proportional gain, and ζ>0,k s >0, α represents exponential gain, and α∈[0,1].

[0055] S35. Through the above steps, the control input can be constructed as:

[0056]

[0057] in,

[0058] S36. Since the speed state is unmeasurable, it is necessary to replace the unavailable speed state in (12) with the speed estimated by the extended high-gain observer. That is, first of all, it can be known that the sliding surface state after replacement is:

[0059]

[0060] Then, the control input is:

[0061]

[0062] Furthermore, the step S4 includes the following sub-steps:

[0063] S41. Using the control input designed in S36 and bringing it into the dynamic model of the multi-agent system, it can be seen that the derivative of the auxiliary variable is:

[0064]

[0065] Furthermore, the sliding surface derivative can be obtained as:

[0066]

[0067] S42, using the derivative information of the corresponding variable obtained in S41, combined with the expanded high-gain observer in step S2:

[0068]

[0069] in, is the speed variable p 2i The estimation error of is the derivative of the velocity estimation error of an individual in a multi-agent system.

[0070] S43. Based on the singular perturbation theory, the extended high-gain observer and the sliding mode controller can be combined to obtain the closed-loop error system described by the fast and slow dynamic equations:

[0071]

[0072] in, is η 2iBy analyzing the fast-slow variable dynamic equation described by (16), it can be seen that if the derivative of the disturbance is adjusted at the time T Δ Then the corresponding quantity in the fast and slow dynamic equations is asymptotically bounded, and its final bound is consistent with the designable parameter ε f Therefore, it can be said that the use of control inputs such as (13) can achieve robust synchronous control.

[0073] Furthermore, the step S5 comprises the following steps:

[0074] The control protocol of step S3 is applied to the dynamic equations of the individuals. When the topological structure is connected, the states of any two individuals can eventually converge to the same state, and the interference and unmeasurable states can be estimated simultaneously, thus completing the design of the robust synchronization control scheme for the multi-agent system.

[0075] Beneficial effects of the present invention: The method of the present invention establishes a second-order dynamic model of individuals in a multi-agent system, designs an improved extended high-gain observer for the model, and then designs a robust synchronization control protocol, constructs a fast-slow dynamic model based on singular perturbations, and designs a robust synchronization controller based on continuous sliding mode theory, so that individuals in the multi-agent system can eventually achieve state synchronization. The method of the present invention focuses on the adverse effects of interference and unmeasurable speed on the multi-agent system, laying a foundation for technicians to simultaneously solve the adverse effects of interference and unmeasurable speed in actual operations, ensuring that the interference estimation error and the speed estimation error tend to a bounded range at the same time, and at the same time, it can analyze the relationship between the convergence of the system synchronization error and the control parameters based on the fast-slow variable state equation, promote parameter adjustment, improve the effectiveness and robustness of the synchronization control scheme, and expand the application scenarios of the algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Figure 1 The present invention is a flow chart of the robust synchronization control method based on the improved expanded high-gain observer.

[0077] Figure 2 This is the communication topology diagram of the multi-agent system in this embodiment.

[0078] Figure 3 It is the position state response diagram of the intelligent agent individual in this embodiment.

[0079] Figure 4 It is the estimation error diagram of the unknown interference in this embodiment.

[0080] Figure 5 Graph showing the estimated error of the unmeasurable state in this embodiment.

[0081] Figure 6 Graph showing the synchronization error response in this embodiment.

[0082] Figure 7 This is the time response diagram of the control input in this embodiment. DETAILED DESCRIPTION

[0083] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

[0084] like Figure 1 As shown, a flow chart of a robust synchronization control method based on an improved extended high-gain observer of the present invention is shown, and the specific steps are as follows:

[0085] S1. Establish the second-order dynamics model of individuals in the multi-agent system in the field of drones and unmanned vehicles, including two channels: position and velocity.

[0086] It includes the following sub-steps:

[0087] S11. The multi-agent system has n individuals, which are connected through an undirected network G. The dynamic model of n second-order individuals is as follows:

[0088]

[0089] in, Represent the position, velocity and control input of the ith individual, represents the real number space, Respectively represent the position derivative and velocity derivative of the ith individual, t represents time, and f i (r i (t),v i (t), t) represents the interference in the control input channel. Due to the limitation of sensors, the speed information of the individual itself cannot be fed back for designing the control protocol. At the same time, an individual in the multi-agent system can only interact with its neighbors (i.e., there is a connection) to complete the information exchange.

[0090] S12. To achieve synchronous control of multi-agent systems, the following definitions are made:

[0091]

[0092] That is, the states of any two individuals in the multi-agent system converge and eventually tend to be consistent. When the conditions shown in (3) are met, the multi-agent system is said to have achieved synchronous control, where r j (t) represents any individual among the n individuals in the multi-agent system except the i-th individual.

[0093] S2. For the second-order dynamic model, an improved extended high-gain observer is designed to estimate the unmeasurable speed and unknown disturbance through a parameter-adjustable dynamic system.

[0094] It includes the following sub-steps:

[0095] S21. First, in order to ensure the effective estimation of interference, it is necessary to make assumptions about the upper bounds of interference and its derivatives.

[0096]

[0097] in, Represents the upper bound of the disturbance and the disturbance derivative, which is a positive number. To control the interference f in the input channel i (r i ,v i ,t), and define the estimated values ​​of individual position, velocity and disturbance in the multi-agent system as The corresponding estimation error is:

[0098]

[0099] in, is the corresponding estimation error, and then for individual i in the multi-agent system, an improved extended high-gain observer is designed in the following form:

[0100]

[0101] in, Represent the corresponding variable estimates The derivative of , where the parameters α1, α2, α3 are selected so that the characteristic polynomial s 3 +α1s 2 +α2s+a3=0 all have negative real parts, where s represents the time domain. In addition, the dynamically adjustable parameter ε is the output state of the following dynamic system:

[0102]

[0103] Where ω represents the system state of the dynamic system, is the derivative of ω, ω(0)=ω0,ω0>ε f , ω0 is a constant, ω(0) represents the initial value of the variable ω, ε f is the preset value of the dynamic parameter, and ε f is a relatively small quantity. In addition, the parameters c1>0, c2>0, 0<λ1<1 and λ2>1, c1 and c2 represent the proportional gain values, and λ1 and λ2 represent the exponential gain values. Furthermore, based on the finite time theory, it can be known that by using the dynamic system shown in (5), the dynamically adjustable parameter ε can reach ε in a fixed time. f , and the corresponding adjustment time is:

[0104]

[0105] Among them, T Δ A variable representing the time it takes for the system to converge.

[0106] S22. When the extended high-gain observer shown in (4) is used, the simultaneous estimation of the unmeasurable state and the unknown interference can be achieved, and the corresponding estimation performance can be adjusted through the dynamic system (5).

[0107] S23. Introduce the coordinate transformation relationship shown below:

[0108]

[0109] Among them, η 1i , η 2i , η 3i Represent the corresponding estimated error variables The conversion amount, combined with formula (4), obtains the following fast-changing time dynamic system:

[0110]

[0111] Among them, η i =[η 1i ,η 2i ,η 3i ] T , representing three variables η 1i ,η 2i ,η 3i The column vector formed, T represents the matrix transpose, To control the interference f in the input channel i The derivative of η represents the system matrix, B η represents the input matrix. For the dynamic system (6), if the system matrix A η is Hurwitz's, and Lyapunov's equation If it holds true, P and Q are matrices of corresponding dimensions, L η is a positive constant. Then we know that the dynamically adjustable parameter ε satisfies:

[0112]

[0113] In the formula, λ max ,λ min are the maximum and minimum eigenvalues ​​of the corresponding matrix respectively, then we can further obtain that the speed estimation error and interference estimation error satisfy:

[0114]

[0115] in, They represent the upper bound functions of the estimated errors of individual speed and disturbance in the multi-agent system, respectively, and there is Furthermore, it can be seen that the improved extended high-gain observer can effectively achieve the effective simultaneous estimation of unmeasurable states and unknown interferences.

[0116] S3, define the synchronization error and design the sliding surface, then compensate the disturbance estimate obtained in step S2 into the controller and replace the unavailable state in the control input with the estimated unmeasurable state, and then design a robust synchronization control protocol,

[0117] It includes the following sub-steps:

[0118] S31, synchronization error is defined as:

[0119]

[0120] Among them, a ij Represents the adjacency matrix in a multi-agent network topology.

[0121] S32. To achieve the convergence of the synchronization error in step S31, the following coordinate transformation relationship is designed:

[0122]

[0123] Among them, p 1i 、p 2i Represent the position error variable and speed variable respectively, and we can get is the speed variable p 2i The derivative of .

[0124] S33. Then, based on the variables defined in (8), the sliding surface of the following form can be constructed:

[0125]

[0126] Among them, s i represents the sliding surface variable, k1 and k2 represent the proportional gain parameters, both of which are positive numbers. Then, the derivative of (9) is:

[0127]

[0128] S34. Based on the classical sliding mode technology, the following sliding mode surface approach law can be constructed:

[0129]

[0130] Among them, ζ, k s represents proportional gain, and ζ>0,k s >0, α represents exponential gain, and α∈[0,1].

[0131] S35. Through the above steps, the control input can be constructed as:

[0132]

[0133] in,

[0134] S36. Since the speed state is unmeasurable, it is necessary to replace the unavailable speed state in (12) with the speed estimated by the extended high-gain observer. That is, first of all, it can be known that the sliding surface state after replacement is:

[0135]

[0136] Then, the control input is:

[0137]

[0138] S4, bring the control protocol design established in step S3 into the original dynamic model to obtain a closed-loop error system, and then define auxiliary variables, and then based on the singular perturbation technology, convert the closed-loop error system into a dynamic equation of fast and slow dynamic coupling to determine the upper bound of the synchronization error,

[0139] It includes the following sub-steps:

[0140] S41. Using the control input designed in S36 and bringing it into the dynamic model of the multi-agent system, it can be seen that the derivative of the auxiliary variable is:

[0141]

[0142] Furthermore, the sliding surface derivative can be obtained as:

[0143]

[0144] S42, with the help of the derivative information of the corresponding variable obtained in S41, combined with the expanded high gain observer in step S2, it can be known that

[0145]

[0146] in, is the speed variable p 2i The estimation error of is the derivative of the velocity estimation error of an individual in a multi-agent system.

[0147] S43. Based on the singular perturbation theory, the extended high-gain observer and the sliding mode controller can be combined to obtain the closed-loop error system described by the fast and slow dynamic equations:

[0148]

[0149] in, is η 2i By analyzing the fast-slow variable dynamic equation described by (16), it can be seen that if the derivative of the disturbance is adjusted at the time T Δ Then the corresponding quantity in the fast and slow dynamic equations is asymptotically bounded, and its final bound is consistent with the designable parameter ε f Therefore, it can be said that the use of control inputs such as (13) can achieve robust synchronous control.

[0150] S5. Apply the control protocol to the second-order dynamics model of the multi-agent system, so that the individuals in the multi-agent system can eventually achieve state synchronization.

[0151] The control protocol of step S3 is applied to the dynamic equations of the individuals. When the topological structure is connected, the states of any two individuals can eventually converge to the same state, and the interference and unmeasurable states can be estimated simultaneously, thus completing the design of the robust synchronization control scheme for the multi-agent system.

[0152] The communication topology diagram of the multi-agent system in this embodiment is as follows Figure 2 As shown, there are 4 individuals, and the unknown disturbance in the individual dynamics model is:

[0153]

[0154] Among them, f1, f2, f3, f4, r1, r2, r3, r4, v1, v2, v3, v4 represent the interference, position and speed of each individual respectively, and the initial values ​​of the individual states are:

[0155] r i (0) = [3.2, -4.6, 5.5, 2.4] T

[0156] v i (0) = [0.8, -0.4, 0.7, 1.3] T

[0157] The control parameters are selected as follows:

[0158] k1=10,k2=2,k s =0.3,ζ=0.3,α=0.3,α1=2,α2=2,α3=2;

[0159] λ1=0.5,λ2=2,c1=5,c1=5,ω0=0.2,ε f =0.05

[0160] visible, Figure 3The position state response of the individual intelligent agent; Figure 4 is the estimation error of the unknown interference; Figure 5 is the estimation error of the unmeasurable state; Figure 6 is the synchronization error response curve of the multi-agent system; Figure 7 is the time response of the control input; it can be seen from the simulation results that the synchronous control of the multi-agent system is realized in the final embodiment, and the estimation error of the unknown interference and unmeasurable state eventually tends to be bounded. The simulation results show that the proposed control scheme can realize the synchronous control of the multi-agent system.

[0161] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.

Claims

1. A robust synchronization control method based on an improved extended high-gain observer, the specific steps comprising the following steps: S1. Establish the second-order dynamics model of individuals in the multi-agent system in the field of drones and unmanned vehicles, including two channels: position and velocity; S2. For the second-order dynamic model, an improved extended high-gain observer is designed to estimate the unmeasurable speed and unknown disturbance through a parameter-adjustable dynamic system; S3, defining the synchronization error and designing the sliding surface, compensating the disturbance estimation value obtained in step S2 into the controller, and at the same time replacing the unavailable state in the control input with the estimated unmeasurable state, thereby designing a robust synchronization control protocol; S4, bringing the control protocol design established in step S3 into the original dynamic model to obtain a closed-loop error system, and then defining auxiliary variables, and then based on the singular perturbation technology, converting the closed-loop error system into a dynamic equation of fast and slow dynamic coupling to determine the upper bound of the synchronization error; S5. Apply the control protocol to the second-order dynamics model of the multi-agent system, so that the individuals in the multi-agent system can eventually achieve state synchronization; The specific process of step S1 is: S11. The multi-agent system has n individuals, which are connected through an undirected network G. The dynamic model of n second-order individuals is as follows: in, Represent the position, velocity and control input of the ith individual, represents the real number space, Respectively represent the position derivative and velocity derivative of the ith individual, t represents time, and f i (r i (t),v i (t),t) represents the disturbance existing in the control input channel; S12. To achieve synchronous control of multi-agent systems, the following definitions are made: Among them, r j (t) represents any individual except the i-th individual in the multi-agent system; The specific process of step S2 is: S21. Make assumptions about the upper bounds of the disturbance and its derivatives. in, Represents the upper bound of the disturbance and the disturbance derivative, which is a positive number. To control the interference f in the input channel i (r i ,v i ,t), and define the estimated values ​​of individual position, velocity and disturbance in the multi-agent system as The corresponding estimation error is: in, is the corresponding estimation error, and then for individual i in the multi-agent system, an improved extended high-gain observer is designed in the following form: in, Represent the corresponding variable estimates The derivative of , where the parameters α1, α2, α3 are selected so that the characteristic polynomial s 3 +α1s 2 +α2s+a3=0 all have negative real parts, where s represents the time domain. In addition, the dynamically adjustable parameter ε is the output state of the following dynamic system: Where ω represents the system state of the dynamic system, is the derivative of ω, ω(0)=ω0,ω0>ε f , ω0 is a constant, ω(0) represents the initial value of the variable ω, ε f is the preset value of the dynamic parameter, and ε f is a relatively small quantity. In addition, the parameters c1>0, c2>0, 0<λ1<1 and λ2>1, c1 and c2 represent proportional gain values, λ1 and λ2 represent exponential gain values. Based on the finite time theory, it can be known that by using the dynamic system shown in (5), the dynamically adjustable parameter ε can reach ε in a fixed time. f , and the corresponding adjustment time is: Among them, T Δ A variable representing the system convergence time; S22. When the extended high-gain observer shown in (4) is used, simultaneous estimation of unmeasurable states and unknown disturbances can be achieved, and the corresponding estimation performance can be adjusted through the dynamic system (5); S23. Introduce the coordinate transformation relationship shown below: Among them, η 1i , η 2i , η 3i Represent the corresponding estimated error variables The conversion amount, combined with formula (4), obtains the following fast-changing time dynamic system: Among them, η i =[η 1i ,η 2i ,η 3i ] T , representing three variables η 1i ,η 2i ,η 3i The column vector formed, T represents the matrix transpose, To control the interference f in the input channel i The derivative of η represents the system matrix, B η represents the input matrix. For the dynamic system (6), if the system matrix A η is Hurwitz's, and Lyapunov's equation If it holds true, P and Q are matrices of corresponding dimensions, L η is a positive constant, then it can be seen that the dynamically adjustable parameter ε satisfies: In the formula, λ max ,λ min are the maximum and minimum eigenvalues ​​of the corresponding matrix respectively, then we can further obtain that the speed estimation error and interference estimation error satisfy: in, They represent the upper bound functions of the estimated errors of individual speed and disturbance in the multi-agent system, respectively, and there is The specific process of step S3 is: S31, synchronization error is defined as: Among them, a ij Represents the adjacency matrix in a multi-agent network topology; S32. Design the following coordinate transformation relationship: Among them, p 1i 、p 2i Represent the position error variable and speed variable respectively, and can be obtained is the speed variable p 2i The derivative of S33. Based on the variables defined in (8), a sliding surface of the following form is constructed: Among them, s i represents the sliding surface variable, k1 and k2 represent the proportional gain parameters, both of which are positive numbers. Then, the derivative of (9) is: S34. Construct the following sliding surface reaching law: Among them, ζ, k s represents proportional gain, and ζ>0,k s >0, α represents exponential gain, and α∈[0,1]; S35, construct the control input as: in, S36, the unavailable velocity state in (12) is replaced by the velocity estimated by the extended high-gain observer, that is, firstly, it can be known that the sliding surface state after replacement is: The control input is then: The specific process of step S4 is: S41. Using the control input designed in S36 and bringing it into the dynamic model of the multi-agent system, it can be seen that the derivative of the auxiliary variable is: The derivative of the sliding surface is obtained as: S42, using the derivative information of the corresponding variable obtained in S41, combined with the expanded high-gain observer in step S2: in, is the speed variable p 2i The estimation error of is the derivative of the velocity estimation error of an individual in a multi-agent system; S43. Based on the singular perturbation theory, the extended high-gain observer and the sliding mode controller can be combined to obtain the closed-loop error system described by the fast and slow dynamic equations: in, is η 2i The derivative of The specific process of step S5 is: The control protocol of step S3 is applied to the dynamic equations of the individuals. When the topological structure is connected, the states of any two individuals can eventually converge to the same state, and the interference and unmeasurable states can be estimated simultaneously, thus completing the design of the robust synchronization control scheme for the multi-agent system.