A multi-agent adaptive cooperative fault-tolerant tracking control method
Through the adaptive collaborative fault-tolerant tracking control method, the distributed sliding mode observer and fuzzy neural network are used to solve the high-precision tracking performance problem of multi-agent systems under external interference and malicious attacks, and the tracking error in fixed time is achieved Convergence and stable performance, suitable for high-precision tracking control of multi-agent systems.
Patent Information
- Application Number
- CN202311391200.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-10-25
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2043-10-25
AI Technical Summary
Multi-agent systems are difficult to maintain high-precision tracking performance under external interference and malicious attacks, especially in the case of actuator failure and limited communication resources, steady-state and transient performance are difficult to guarantee.
A multi-agent adaptive collaborative fault-tolerant tracking control method is proposed. By constructing a system model under interference or attack, estimating the leader's state using a distributed sliding mode observer, calculating synchronous tracking errors, and fuzzy neural networks and Nuss Baum function designs an adaptive control law, converts the preset performance inequality into bounded constraint error, and realizes tracking error convergence within a fixed time.
The tracking error of the multi-agent system is converged into a small neighborhood within a fixed time, effectively identifying and resisting malicious attacks, ensuring stable tracking performance, and is suitable for nonlinear multi-agent systems with actuator failures and limited communication bandwidth.
Smart Images

Figure CN117687434B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of multi-agent system control, and in particular to a multi-agent adaptive collaborative fault-tolerant tracking control method. Background Art
[0002] The research on multi-agent cooperative tracking control is an important topic in specific military tasks such as satellite group formation navigation and positioning, drone group formation coordinated attack, and missile group attack on time-sensitive targets. The good tracking performance of multi-agent system (MAS) depends on accurate system model, strong fault tolerance and anti-interference ability. In practical engineering, high-precision tracking performance is limited by three factors: the accuracy of MAS model; the steady-state performance and transient performance of MAS; and the limited system communication resource constraints.
[0003] In order to solve the problem of uncertain system models and unknown fault models, current research mainly includes three aspects: 1) Nonlinear uncertainty in the system model. For unknown control gains, the Nussbaum function is currently a good solution for searching for uncertainty control directions. 2) The coupled uncertainty of the system model is subject to external interference and attack. Many schemes use different neural network methods to estimate the uncertainty of the system model. 3) Uncertainty of the fault model. At present, most of the research on system failures focuses on actuator failures and sensor failures. In actual engineering, it is necessary to ensure the transient performance and steady-state performance of MAS under external interference. In order to solve this problem, many scholars have studied preset performance control (PPC).
[0004] Many solutions have been proposed for high-precision tracking performance control of MAS, but they still face the following problems: (i) MAS is subject to external interference and malicious attacks, which makes it difficult to know the state of the leader and the actuator fault model. (ii) How to solve the problem of comprehensive unknown control direction of high-order MAS. (iii) How to ensure steady-state performance and transient performance under external interference within a fixed time. Summary of the invention
[0005] The purpose of the present invention is to overcome the defects of the prior art and solve the problem of high-precision tracking performance control of multiple agents.
[0006] To achieve the above objectives, the present invention proposes a multi-agent adaptive collaborative fault-tolerant tracking control method, comprising the following steps:
[0007] S1. Construct a multi-agent system model under external interference or malicious attack. The specific form is: Where i represents the i-th agent, i=1,...,N, For the qth i state variables, 1≤q i ≤ni -1, is a bounded unknown nonlinear function, is the unknown interference function, is the unknown time-varying gain, y i is the output signal, To quantify the input is the dead zone width of the quantized signal, σ i =(1-θ i ) / (1+θ i ) is the quantized density, θ i ∈(0,1), is the actuator fault model, ρ i (t) is a partial loss of effectiveness failure, is the bias fault signal, r i (t) is the bounded nonlinear fault caused by malicious attack, and is the bounded energy function, s i (t)∈[0,1] obeys the Bernoulli equation.
[0008] S2. Establish a distributed sliding mode observer to estimate the unknown leader state information. The specific form is: in For the status information of the pilot The estimate, is the estimation error. are observer parameters.
[0009] S3. Calculate the synchronous tracking error, the specific form is: where η i,1 is the estimated value of the observer, b i is the tracking control coefficient, δ i and δ j is the model reconstruction coefficient. In the directed communication graph of MAS, A=[a ij ]∈R N×N is the adjacency matrix, D = diag{D i} is the degree matrix, where L=DA is the Laplace matrix.
[0010] S4. Set the preset performance inequality constraint error Transformed into bounded constraint error E i,1 The specific process is: design a fixed time preset performance function where k i,h,0 , k i,h,∞ , τ i,h , T i,h is the design parameter, and the error conversion function is λ λi,h is the design parameter, i,h Differentiate, and we get In the formula
[0011] S5. Determine whether the calculation condition h≥2 is satisfied at this time, indicating the order of h calculation. If not, execute S6 and S8. If yes, execute S7 and S8.
[0012] S6. Define the tracking error as in Right i,1 Differentiation, design virtual control law as α i,1 =μ i,1 N(ζ i,1 )φ i,1 , where μ i,1 is the design parameter, is the Nussbaum function, m i,1 is the design parameter, The adaptive control law is
[0013]
[0014] S7, z i,h Differentiate, and we get Using a second-order robust tracking differentiator, The specific form is in They are α i,h-1 , The estimated value of Bounded estimation error, ι i,h , ι i,h+1 To design parameters, fuzzy neural network is used to estimate in is the ideal weight vector, Ψ i,h (X i,h ) is a Gaussian function, and the designed virtual control law is α i,h =μ i,h N(ζ i,h )φ i,h , where μ i,h is the design parameter, is the Nussbaum function, m i,h is the design parameter, and the adaptive control law is
[0015]
[0016] S8. h=h+1, execute step S5.
[0017] S9, when h = n i +1, and the actual control input is
[0018] Furthermore, it is proved that the tracking error of the multi-agent system using this control method can converge to a small neighborhood within a fixed time.
[0019] Furthermore, a computer-readable storage device is provided, wherein the storage device stores a computer program, and wherein the computer program, when executed, implements a multi-agent adaptive collaborative fault-tolerant tracking control method.
[0020] Compared with the prior art, the present invention has the following technical effects: for nonlinear multi-agent systems with actuator failures, limited communication bandwidth and unknown control gain problems, a new adaptive collaborative fault-tolerant tracking control method is proposed, and a distributed high-order sliding mode observer is used to estimate the state of the leader. At the same time, the observer can also identify some unknown and uncertain malicious attacks; a new tracking error conversion mechanism is proposed, and the mechanism is applied to convert the constrained error into an unconstrained tracking error, and a fuzzy logic system is used to estimate the unknown nonlinear function in the multi-agent system; a fixed-time preset performance function is designed to achieve stable tracking performance; it is proved that the tracking error of the multi-agent system can converge to a small neighborhood within a fixed time. BRIEF DESCRIPTION OF THE DRAWINGS
[0021] Figure 1 This is a flow chart of a multi-agent adaptive collaborative fault-tolerant tracking control method of the present invention.
[0022] Figure 2 This is the multi-agent communication topology diagram of the present invention.
[0023] Figure 3 It is a reference trajectory output curve diagram of the present invention under preset performance.
[0024] Figure 4 It is a tracking error curve diagram of the follower of the present invention.
[0025] Figure 5 It is the error curve diagram of the leader state estimation and high-order sliding mode observer of the present invention.
[0026] Figure 6 It is a comparison curve diagram of the control input quantization signal of the present invention and the normal control input signal.
[0027] Figure 7 It is a fuzzy estimation curve diagram of unknown interference of the present invention.
[0028] Figure 8 It is a tracking error curve diagram of the multi-agent tracking trajectory relative to the reference signal in the case of actuator failure of the present invention. DETAILED DESCRIPTION
[0029] In order to further illustrate the features of the present invention, please refer to the following detailed description and drawings of the present invention. The drawings are for reference and illustration only and are not intended to limit the scope of protection of the present invention. The following is a detailed description of the specific implementation of the present invention in conjunction with the drawings:
[0030] like Figure 1 As shown, this embodiment discloses a multi-agent adaptive collaborative fault-tolerant tracking control scheme for multi-agent control, including the following steps:
[0031] S1. Construct a multi-agent system model under external interference or malicious attack. The specific form is: Where i represents the i-th agent, i=1,...,N, For the qth i state variables, 1≤q i ≤n i -1, is a bounded unknown nonlinear function, is the unknown interference function, is the unknown time-varying gain, y i is the output signal, To quantify the input
[0032] is the dead zone width of the quantized signal, σ i =(1-θ i ) / (1+θ i ) is the quantized density, θ i ∈(0,1), is the actuator fault model, ρ i (t) is a partial loss of effectiveness failure, is the bias fault signal, r i (t) is the bounded nonlinear fault caused by malicious attack, and is the bounded energy function, s i (t)∈[0,1] obeys the Bernoulli equation.
[0033] S2. Establish a distributed sliding mode observer to estimate the unknown leader state information. The specific form is: in To provide information about the status of the navigator, The estimate, is the estimation error. are observer parameters.
[0034] S3. Calculate the synchronous tracking error, the specific form is: where η i,1 is the estimated value of the observer, b i is the tracking control coefficient, δ i and δ j is the model reconstruction coefficient. In the directed communication graph of MAS, A=[ai j ]∈R N×N is the adjacency matrix, D = diag{D i} is the degree matrix, where L=DA is the Laplace matrix.
[0035] S4. Set the preset performance inequality constraint error Transformed into bounded constraint error E i,1 The specific process is: design a fixed time preset performance function where k i,h,0 , k i,h,∞ , τ i,h , T i,h is the design parameter, and the error conversion function is λ i,h is the design parameter, i,h Differentiate, and we get In the formula
[0036] S5. Determine whether the calculation condition h≥2 is satisfied at this time, indicating the order of h calculation. If not, execute S6 and S8. If yes, execute S7 and S8.
[0037] S6. Define the tracking error as in Right i,1 Differentiation, design virtual control law as α i,1 =μ i,1 N(ζ i , 1)φ i,1 , where μ i,1 is the design parameter, is the Nussbaum function, m i,1 is the design parameter, The adaptive control law is
[0038]
[0039] S7, Z i,hDifferentiate, and we get Using a second-order robust tracking differentiator, The specific form is in They are α i,h-1 , The estimated value of Bounded estimation error, ι i,h , ι i,h+1 To design parameters, fuzzy neural network is used to estimate in is the ideal weight vector, Ψ i,h (X i,h ) is a Gaussian function, and the designed virtual control law is α i,h =μ i,h N(ζ i,h )φ i,h , where μ i,h is the design parameter, is the Nussbaum function, m i,h is the design parameter, and the adaptive control law is
[0040]
[0041] S8. h=h+1, execute step S5.
[0042] S9, when h = n i +1, and the actual control input is
[0043] Furthermore, it is proved that the tracking error of the multi-agent system using this control method can converge to a small neighborhood within a fixed time. The specific process is:
[0044] The synchronous tracking error is z i,1 Differentiation, there is
[0045]
[0046] make At the same time Substituting into, formula (1) can be simplified to
[0047]
[0048] Select the Lyapunov function as
[0049]
[0050] Differentiating equation (3) yields:
[0051]
[0052] make And the fuzzy neural network is used to estimate G i,1 , can be obtained
[0053]
[0054]
[0055] M i,J d i,1 -A ij d j,1 +ε i,1 ≤Θ i,j (7)
[0056] There is a lemma that for any χ∈R, the following inequality holds
[0057]
[0058] Based on the lemma of formula (8), and by introducing the adaptive control law, formula (4) can be rearranged as
[0059]
[0060] When 2≤h≤n i When , there is an error equation
[0061] Select the Lyapunov function as
[0062]
[0063] The quantized input signal equation in the system can be converted into
[0064] Q(u i )=h(u i ) i +Δ(u i ) (11)
[0065] If Q(u i )=0, then h(u i )=1,Δ(u i )=-u i ; If Q(u i )≠0, then h(u i )=Q(u i ) / u i , Δ(u i )=0.
[0066] Differentiating equation (10) yields
[0067]
[0068] The fuzzy neural network estimation is the ideal weight vector, Ψ i,h (X i,h ) is a Gaussian function,
[0069] Then there is
[0070]
[0071] Based on the lemma of equation (8), and by introducing the adaptive control law, equation (12) can be organized as
[0072]
[0073] Consider a multi-agent system consisting of a leader and N-1 followers, and choose the Lyapunov function as
[0074]
[0075] Differentiate equation (15), substitute equation (9) and equation (14) and simplify to obtain
[0076]
[0077] in
[0078] Multiply both ends of formula (16) by And at time [0,t f ) The integral is
[0079]
[0080] There is a formula where ζ i,l , E i,l All are bounded.
[0081] Further, according to Available
[0082]
[0083]
[0084] From the above formula, we can see that the transformed tracking error can converge to the specified range within a fixed time. The proof is complete.
[0085] A person of ordinary skill in the art can understand that all or part of the steps of implementing the above method embodiment can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiment; and the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk, etc., various media that can store program codes.
[0086] The following is a simulation example.
[0087] Consider Figure 2 The multi-agent system consists of a leader and four followers. The second-order dynamics model of the multi-agent system is shown in
[0088]
[0089] where d i,1 (x i,1 )=x i,1 sin(0.5x i,1 ), d i,2 (x i,1 x i,2 )=1+x i,1 sin(0.5x i,1 x i,2 ) is external interference, and the initial state is designed to be [x i,1 (0), x i,2 (0)] T =[1.25, 0.25] T , and the adjacency matrix of the topological graph is The degree matrix is The Laplace matrix is
[0090] The design preset performance function is The initial value of the distributed high-order sliding mode observer is η i,1 =[0.183, 0] T , adaptive control law, the initial value of external disturbance and actuator fault is 0. In order to reduce the chattering phenomenon caused by the sign function in the controller design process, the hyperbolic tangent function is used for smoothing.
[0091] The reference tracking signal is y L = sin(t)+sin(1.5t), the center value and width parameters of the designed fuzzy neural network are The initial state of the 4 followers is μ i =0deg,α i =1.83deg,β i (0) = 0 degrees, pi (0) = q i (0) = r i (0) = 0 degrees.
[0092] Finally, the reference trajectory output under the preset performance is obtained as follows Figure 3 As shown in Figure 2, the tracking error of the follower is Figure 4 As shown in , it can be seen that under the action of the designed controller, the tracking error of the follower is within the allowable range, the intelligent agent tracks the desired position well, and achieves the established formation and stable state. The error curves of the leader state estimation and the high-order sliding mode observer are shown in Figure 5 As shown, the comparison between the control input quantization signal and the normal control input signal is Figure 6 As shown, the fuzzy estimation curve of unknown interference is as follows Figure 7 shown.
[0093] To further illustrate the effectiveness of this method in dealing with actuator failure, the actuator failure state is added to obtain the tracking error curve of the multi-agent tracking trajectory relative to the reference signal under the actuator failure condition. Figure 8 As shown, it can be seen that when the follower encounters an actuator fault, the tracking error can still be kept in a very small area under the action of the designed controller.
[0094] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims
1. A multi-agent adaptive collaborative fault-tolerant tracking control method, characterized in that: The steps include: S1. Construct a multi-agent system model under external interference or malicious attack; S2, establish a distributed sliding mode observer to estimate the unknown leader state information; S3, calculating synchronous tracking error; S4. Set the preset performance inequality constraint error Transformed into bounded constraint error E i,1 ; S5, determine whether the calculation condition h≥2 is met at this time, where h represents the order of calculation. If not, execute S6 and S8; if yes, execute S7 and S8; S6. Define the tracking error as in Right i,1 Differentiation, design virtual control law as α i,1 =μ i,1 N(ζ i,1 )φ i,1 , where μ i,1 is the design parameter, is the Nussbaum function, m i,1 is the design parameter, S7, z i,h Differentiate, and we get A second-order robust tracking differentiator is used to Make estimates; S8, when h=h+1, execute step S5; S9, when h = n i +1, and the actual control input is Among them, in step S1, the multi-agent system model is specifically in the form of: Where i represents the i-th agent, i=1,...,N, For the qth i state variables, 1≤q i ≤n i -1, is a bounded unknown nonlinear function, is the unknown interference function, is the unknown time-varying gain, y i is the output signal, To quantize the input signal, is the dead zone width of the quantized signal, σ i =(1-θ i ) / (1+θ i ) is the quantized density, θ i ∈(0,1), is the actuator fault model, ρ i (t) is a partial loss of effectiveness failure, is the bias fault signal, r i (t) is the bounded nonlinear fault caused by malicious attack, and is the bounded energy function, s i (t)∈[0,1] obeys the Bernoulli equation; Among them, in step S2, the distributed sliding mode observer estimates the unknown leader state information, and the specific form is: in For the status information of the pilot The estimate, is the estimation error; is the observer parameter; Among them, in step S3, the synchronous tracking error is specifically in the form of: Where ηi,1 is the estimated value of the observer, bi is the tracking control coefficient, δ i and δ j is the model reconstruction coefficient. In the directed communication graph of MAS, A=[a ij ]∈R N×N is the adjacency matrix, D = diag{D i } is the degree matrix, where L = DA is the Laplace matrix; Among them, in step S4, the fixed time preset performance function is designed as: in, k i,h,0 , k i,h,∞ , τ i,h , T i,h is the design parameter, and the error conversion function is λ i,h is the design parameter, i,h Differentiate, and we get In the formula Among them, in step S6, the adaptive control law is: Among them, in step S7, the specific form is: in They are α i,h-1 , The estimated value of Bounded estimation error, ι i,h , ι i,h+1 To design parameters, fuzzy neural network is used to estimate in is the ideal weight vector, Ψ i,h (X i,h ) is a Gaussian function, and the designed virtual control law is α i,h =μ i,h N(ζ i,h )φ i,h , where μ i,h is the design parameter, is the Nussbaum function, m i,h is the design parameter.
2. A multi-agent adaptive collaborative fault-tolerant tracking control method according to claim 1, characterized in that: The adaptive control law is:
3. A multi-agent adaptive collaborative fault-tolerant tracking control method according to claim 1 or 2, characterized in that: The tracking error of the multi-agent system converges to a small neighborhood within a fixed time. The specific process is: The synchronous tracking error is z i,1 Differentiation, there is make At the same time Substitute it in and simplify it to: Select the Lyapunov function as: To find the differential, make And the fuzzy neural network is used to estimate G i,1 ,get M i,J d i,1 -A ij d j,1 +ε i,1 ≤Θ i,1 For any χ∈R, the following inequality holds At the same time, the adaptive control law is introduced as follows: When 2≤h≤n i When , there is an error equation 4. A multi-agent adaptive collaborative fault-tolerant tracking control method according to claim 3, characterized in that: Select the Lyapunov function as: The quantized input signal equation in the system is converted to: Q(in i )=h(u i )in i +Δ(in i ) Among them, if Q(u i ) = 0, then h(u i ) = 1, Δ(u i ) = -u i ; if Q(u i ) ≠ 0, then, h(in i )=Q(u i ) / in i , Δ(u i )=0; The differential is: Among them, the fuzzy neural network is used to estimate is the ideal weight vector, Ψ i,h (X i,h ) is a Gaussian function, Then we have: At the same time, the adaptive control law is introduced and sorted into:
