Multi-agent system fault detection method and device influenced by channel fading

Abstract

The invention provides a fault detection method and device for a multi-agent system influenced by channel fading, and relates to the technical field of multi-agent control. The method comprises the steps that the multi-agent system is modeled into a linear variable parameter system, external disturbance and fault signals are fully considered, a foundation is laid for subsequent operation, a fault detection observer and a consistency controller are constructed, and the fault detection accuracy is improved. The former monitors faults, the latter unifies agent behaviors, introduces an event trigger mechanism to reduce resource consumption, improves system collaboration and operation efficiency, calculates and detects observer gain and consistency controller gain according to specific conditions, updates control input based on the gain, calculates residual signals, and provides a basis for fault judgment; the fault is judged by using the residual error evaluation function and the threshold value, the method is intuitive and efficient, an alarm can be given in time, the system performance, safety and reliability are guaranteed, and the stability and fault detection capability of the multi-agent system in a channel fading environment are improved.

Description

Technical Field

[0001] The present invention mainly relates to the field of multi-agent control technology, and in particular to a method and device for detecting faults in a multi-agent system affected by channel fading. Background Art

[0002] Multi-agent systems are a current research hotspot and have been widely used in both industrial and military fields. As a distributed system, multi-agent systems consist of multiple agents with independent perception, decision-making, and communication capabilities, completing control tasks by exchanging information between adjacent agents. Compared to single agents, multi-agent systems offer significant advantages in efficiency, cost, and scalability. In modern network communications, the ubiquitous phenomenon of channel fading can lead to decreased system performance, increased resource consumption, and increased susceptibility to failure. In practical applications, complex multi-agent systems are affected by channel fading. If certain faults cannot be detected and corrected in a timely manner, they can affect the system's performance, safety, and reliability, posing threats to the environment and personal safety. Therefore, studying fault detection methods for multi-agent systems affected by channel fading has important theoretical significance and practical value. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to address the deficiencies of the existing technology and provide a method and device for detecting faults in a multi-agent system affected by channel fading.

[0004] The present invention solves the above-mentioned technical problem with the following technical solution: A method for detecting faults in a multi-agent system affected by channel fading comprises the following steps:

[0005] S1. Describe the multi-agent system in the form of a linear variable parameter system model;

[0006] S2. Construct a fault detection observer for detecting faults in a multi-agent system based on a linear variable parameter system model and a consistency controller for unifying the behaviors of each agent in the multi-agent system;

[0007] S3. Calculate the gain of the fault detection observer based on the asymptotic stability theory of the error system and the sufficient conditions for the existence of the gain of the fault detection observer, and calculate the gain of the consistency controller based on the stochastic stability theory of the closed-loop system and the sufficient conditions for the existence of the gain of the consistency controller;

[0008] S4. updating the input of the linear variable parameter system model using the fault detection observer gain and the consistency controller gain, updating the output based on the input of the linear variable parameter system model, and performing a difference calculation between the updated output of the linear variable parameter system model and the output of the fault detection observer to obtain a residual signal value;

[0009] S5. Perform fault judgment on the residual signal value based on the residual evaluation mechanism to determine whether the multi-agent system has failed.

[0010] Another technical solution of the present invention to solve the above technical problem is as follows: A multi-agent system fault detection device affected by channel fading, comprising:

[0011] System state description module, used to describe the multi-agent system in the form of a linear variable parameter system model;

[0012] An observer and controller building module, which is used to build a fault detection observer for detecting faults in a multi-agent system and a consensus controller for unifying the behaviors of each agent in the multi-agent system based on a linear variable parameter system model;

[0013] A gain calculation module is used to calculate the gain of the fault detection observer based on the asymptotic stability theory of the error system and the sufficient conditions for the existence of the gain of the fault detection observer, and to calculate the gain of the consistency controller based on the stochastic stability theory of the closed-loop system and the sufficient conditions for the existence of the gain of the consistency controller;

[0014] A state update module is used to update the input of the multi-agent system through the gain of the fault detection observer and the gain of the consistency controller, update the output based on the input of the multi-agent system, and calculate the difference between the output of the multi-agent system and the output of the fault detection observer to obtain a residual signal value;

[0015] The fault analysis module is used to perform fault judgment on the residual signal value based on the residual evaluation mechanism, so as to obtain the result of whether the multi-agent system has a fault.

[0016] Another technical solution of the present invention to solve the above-mentioned technical problem is as follows: A multi-agent system fault detection device affected by channel fading includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the multi-agent system fault detection method affected by channel fading as described above is implemented.

[0017] Another technical solution of the present invention to solve the above technical problems is as follows: a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the multi-agent system fault detection method affected by channel fading as described above.

[0018] The beneficial effects of the present invention are: the multi-agent system is modeled as a linear variable parameter system, which can accurately reflect the dynamic characteristics of the system, fully consider external disturbances and fault signals, and lay a solid foundation for subsequent operations. Secondly, a fault detection observer and a consistency controller are constructed for fault monitoring and unified agent behavior, and an event triggering mechanism is introduced to reduce resource consumption, improve system coordination and operating efficiency, and calculate two gains according to specific conditions. With the help of theoretical derivation and tool solution, the performance of the controller and observer is guaranteed; based on the gain, the control input is updated and the residual signal is calculated to provide a key basis for fault judgment. Finally, the residual evaluation function and threshold are used to judge faults. The method is intuitive and efficient, and can detect faults and alarm in time to ensure system performance, safety and reliability, and overall improve the stability and fault detection capability of the multi-agent system in a channel fading environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 is a schematic flow chart of a multi-agent system fault detection method;

[0020] Figure 2 Design flow chart for consistency controller;

[0021] Figure 3 This is the communication topology diagram of a vertical take-off and landing aircraft system;

[0022] Figure 4 It is the residual evaluation function diagram of each agent in the linear variable parameter multi-agent system;

[0023] Figure 5 It is the state component curve diagram of each intelligent agent under the control protocol;

[0024] Figure 6 Fault detection flow chart of the multi-agent system fault detection method

[0025] Figure 7 This is the functional module block diagram of the multi-agent system fault detection method. DETAILED DESCRIPTION

[0026] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only used to explain the present invention and are not used to limit the scope of the present invention.

[0027] Example 1: Figure 1 、 Figure 6 As shown, an embodiment of the present invention provides a method for detecting faults in a multi-agent system affected by channel fading, comprising the following steps:

[0028] S1. Describe the multi-agent system in the form of a linear variable parameter system model;

[0029] S2. Construct a fault detection observer for detecting faults in a multi-agent system based on a linear variable parameter system model and a consistency controller for unifying the behaviors of each agent in the multi-agent system;

[0030] S3. Calculate the gain of the fault detection observer based on the asymptotic stability theory of the error system and the sufficient conditions for the existence of the gain of the fault detection observer, and calculate the gain of the consistency controller based on the stochastic stability theory of the closed-loop system and the sufficient conditions for the existence of the gain of the consistency controller;

[0031] S4. updating the input of the linear variable parameter system model using the fault detection observer gain and the consistency controller gain, updating the output based on the input of the linear variable parameter system model, and performing a difference calculation between the updated output of the linear variable parameter system model and the output of the fault detection observer to obtain a residual signal value;

[0032] S5. Perform fault judgment on the residual signal value based on the residual evaluation mechanism to determine whether the multi-agent system has failed.

[0033] In the above examples, the multi-agent system is described as a linear variable parameter system model, which accurately depicts the relationship between the state of each agent in the system and changes in parameters over time. This model not only considers the time-varying characteristics of the multi-agent system matrix, but also accounts for external disturbances and fault signals to be detected. This provides a solid foundation for subsequent accurate analysis of system behavior and the design of fault detection and control strategies, improving the model's adaptability to complex real-world scenarios.

[0034] A fault detection observer can be constructed to estimate the system state in real time. By comparing it with the actual output and generating a residual signal, it can effectively monitor system faults. An event-triggered mechanism is introduced to construct a consistency controller, which can trigger control signal updates based on changes in agent states. This reduces unnecessary communication and computing resource consumption, improves system efficiency, and ensures the consistency of the behavior of each agent in the multi-agent system, enhancing the overall system collaboration.

[0035] Based on sufficient conditions for the asymptotic stability of the error system and the stochastic stability of the closed-loop system, the gains of the fault detection observer and the consistency controller are calculated. The calculation conditions are derived using theories such as the Lyapunov function and Schur's complement lemma, and solved using MATLAB's LMI toolbox. This provides a solid theoretical foundation and reliable calculation method for the gain calculation, ensuring the performance of the fault detection observer and consistency controller, improving system stability and fault detection accuracy.

[0036] By updating the control input, the new output of the fault detection observer is obtained and the residual signal is calculated. This process, based on the previously calculated gain, ensures the rationality of the system state update and the validity of the residual signal, providing key data for subsequent accurate judgment of whether the system has a fault.

[0037] Based on the residual evaluation mechanism, a residual evaluation function and threshold are defined, and system failures are determined by comparing the two. This simple, intuitive, and effective approach can quickly and accurately identify system failures and generate timely alarm information, enabling timely response measures and minimizing the impact of failures on system performance, safety, and reliability.

[0038] Preferably, if Figure 6 As shown, in S1, the multi-agent system is described as a linear variable parameter system model, including:

[0039] Determine the number of agents in the multi-agent system as N, and the linear variable parameter system model for the i-th agent is expressed as:

[0040]

[0041] in, is the state vector, is the output vector, is the control input, i∈N, For external disturbances, is the fault signal to be detected in the multi-agent system, and the multi-agent system matrix C and D ω and D f are all known constant matrices, ρ(k) is the scheduling vector, A(ρ(k)), B u (ρ(k)),B ω (ρ(k)) and B f (ρ(k)) is a time-varying matrix that depends on the scheduling vector ρ(k). The scheduling parameter ρ(k) satisfies the following conditions:

[0042]

[0043] Among them, ρ s and is a known scalar.

[0044] It should be understood that for a multi-agent system consisting of N agents, the state space model of the i-th agent is described by a linear variable parameter system model as equation (1). Considering that there are many factors that affect signal quality during signal propagation, the controller receives the signal as follows:

[0045]

[0046] Among them:k =min{L,k}, L represents the number of paths; α s (k)(s=0,1,...,l k ) are channel fading coefficients, which are random variables, independent of each other and have a mean of The variance is The same Gaussian distribution.

[0047] In the above embodiment, the multi-agent system is represented as a linear variable parameter system model. The model can fully consider the time-varying characteristics of the parameters in the system, as well as external disturbances and fault signals to be detected, accurately describe the dynamic behavior of each agent, and provide an accurate model basis for the subsequent construction of fault detection observers and consistency controllers, thereby improving the model's ability to describe complex multi-agent systems.

[0048] Preferably, in S2, constructing a fault detection observer for detecting faults in a multi-agent system and a consistency controller for unifying the behaviors of each agent in the multi-agent system based on a linear variable parameter system model includes:

[0049] Assume that the estimated state of the fault detection observer is The output of the fault detection observer is The residual signal is r i (k) Constructing a fault detection observer based on the time-varying matrix and the constant matrix in the linear variable parameter system model, wherein the fault detection observer is:

[0050]

[0051] Among them, H a (ρ(k)) and H b (ρ(k)) is the observer gain;

[0052] Assume that the control input of agent i at time k is u i (k) Based on the scheduling vector in the linear variable parameter system model and introducing an event trigger mechanism, a consistency controller is constructed. The consistency controller is:

[0053]

[0054] in, The trigger condition of the event trigger mechanism, that is, the trigger moment When is the current broadcast state of agent i, agent j is the neighbor agent of agent i, is the state information transmitted from the neighboring agent, α s (k)(s=0,1,...,l k ) is the channel fading coefficient, and K(ρ(k)) is the consistency controller gain.

[0055] It should be understood that, considering that communication channels are often affected by fading behavior and bandwidth constraints, in order to mitigate the adverse effects of packet loss and delay, an adaptive event triggering mechanism (12) is introduced, and the triggering conditions are as follows:

[0056]

[0057] in: Ω(ρ(k)) is a positive definite matrix associated with the scheduling vector ρ(k). σ i (k) is the trigger threshold determined according to the following adaptive rules:

[0058]

[0059] Where: i (0) = σ i,0 ∈(0,1) is the given initial condition.

[0060] Design the fault detection observer (2) and the consistency controller (3).

[0061] Define the estimated error vector:

[0062]

[0063] Considering Equations (1), (2) and (14) comprehensively, the error dynamic system can be obtained as follows:

[0064]

[0065] Among them, ε(k)=col{ε1(k),ε2(k),...,ε N (k)},ω(k)=col{ω1(k),ω2(k),...,ω N (k)}, f(k)=col{f1(k),f2(k),...,f N (k)}, r(k)=col{r1(k),r2(k),...,r N (k)},

[0066]

[0067] By comprehensively considering equations (2) and (3), we can obtain the dynamic change process of the consistency error variable after linear transformation (Equation (17)) and the dynamic change process of the auxiliary variable η(k) (Equation (18):

[0068]

[0069] in, δ(k)=col{δ1(k),δ2(k),...,δ N (k)}, ξ(k)=col{χ(k-1), χ(k-2),..., χ(kL)}, θ(k)=col{ε(k),ω(k),f(k)},

[0070]

[0071] Through the design of the above fault detection observer and consistency controller, the main task of designing the fault detection observer is to make the following conditions hold:

[0072] (i) For any non-zero initial state, the error system (16) is asymptotically stable;

[0073] (ii)||r(k)||≤γ1||ω(k)||,||r(k)||≥β||f(k)||, where γ1 represents the suppression level of the residual to interference and β represents the sensitivity of the residual to faults.

[0074] The main task of designing a consistency controller is to ensure that the following conditions hold:

[0075] (i) For any non-zero initial state, the closed-loop system is asymptotically stable when θ(k) is zero;

[0076] (ii) Under zero initial conditions, for any nonzero θ(k), the state variable χ(k) satisfies E{||χ(k)||}≤γ2E{||θ(k)||}, where γ2 represents H ∞ Performance indicators.

[0077] In the above embodiment, a fault detection observer and a consistency controller are designed. The two cooperate with each other to detect faults and ensure the consistency of the multi-agent system, thereby improving the overall performance of the system. The introduction of an event trigger mechanism reduces unnecessary communication and computing burdens, thereby improving the efficiency and response speed of the system. Through the design of the fault detection observer and the consistency controller, the robustness of the system under channel fading and external disturbances is enhanced.

[0078] Preferably, in S3, calculating the gain of the fault detection observer in the current state based on sufficient conditions for the asymptotic stability of the error system and the existence of the gain of the fault detection observer includes:

[0079] Introducing an error system

[0080] Among them, ε(k)=col{ε1(k),ε2(k),...,ε N (k)},ω(k)=col{ω1(k),ω2(k),...,ω N (k)};

[0081] An error system is introduced, and a Lyapunov function is constructed based on the error system. The Lyapunov function is:

[0082]

[0083] According to the Lyapunov function, we can deduce:

[0084]

[0085] in,

[0086] When the first inequality (21) holds, the error system is asymptotically stable and satisfies H ∞ Disturbance suppression level, the first inequality (21) is:

[0087]

[0088] By applying Schur's complement lemma, we obtain the second inequality (22):

[0089]

[0090] Multiply both sides of the second inequality (22) by U and U respectively. T ,make H a (ρ(k))=X -1 Y ah , H b (ρ(k))=X -1 Y bh , while introducing the condition So the third inequality (4) holds true:

[0091]

[0092] in,

[0093]

[0094] The third inequality (4) is a sufficient condition for the asymptotic stability of the error system and the existence of the fault detection observer gain, Q h and Q t is an unknown positive definite matrix, X, Y ah、Y bh (h, t=1, 2, ..., λ) is the unknown matrix, γ1 and β are unknown scalars, and all other variables are known;

[0095] Use the LMI toolbox in MATLAB to solve the third inequality. If there are scalars γ1>0, β>0 and a positive definite matrix Q h , Q t , and matrices X, Y ah 、Y bh (h, t=1,2,...,λ), so that the third inequality holds, that is, the error system is asymptotically stable and satisfies the joint H ∞ / H - performance index, and thereby obtain the fault detection observer gain, which is expressed as:

[0096]

[0097] If there is no feasible solution to the third inequality, the fault detection observer gain cannot be obtained.

[0098] In the above example, by constructing a Lyapunov function and incorporating the Schur's complement lemma, sufficient conditions for the asymptotic stability of the error system and the existence of the fault detection observer gain were derived. Solving these conditions using the MATLAB LMI toolbox enabled accurate calculation of the fault detection observer gain. This method provides a theoretical basis and a feasible computational approach for the design of the fault detection observer, ensuring the accuracy and effectiveness of fault detection.

[0099] Preferably, if Figure 2 As shown, in S3, based on the sufficient conditions for the existence of random stability of the closed-loop system and the gain of the consistency controller, the gain of the consistency controller in the current state is calculated, including:

[0100] Construct a Lyapunov function:

[0101] V(χ(k))=V1(k)+V2(k) (17)

[0102] in,

[0103] Based on the random stability theory, the stability condition expression (24) is derived:

[0104]

[0105] The following results are derived:

[0106]

[0107] Rewrite q(k) and g(k) from concentrated variable form to component form:

[0108] q(k)=col{q1(k),q2(k),...,q N (k)},

[0109] g(k)=col{q(k-1),q(k-2),...,q(kL)},

[0110] It should be understood that the former is the formula derivation process written in a centralized form, and the latter is written as a set of column vectors, indicating that the centralized variables are composed of components.

[0111] in,

[0112] Σ0=diag{σ 10 ,σ 20 ,...,σ N0},

[0113]

[0114] Based on the closed-loop system in H ∞ It is stochastically stable under the norm constraint and must satisfy the fourth inequality (27):

[0115]

[0116] Using Schur's complement lemma, we construct the fifth inequality (28):

[0117]

[0118] in,

[0119] Defining variables

[0120] Multiply both sides of the fifth inequality by J T With J, we get the sixth inequality (6), and introduce the condition So the sixth inequality holds true, the sixth inequality (6) is:

[0121]

[0122] in,

[0123]

[0124]

[0125] Among them, the sixth inequality (6) is a sufficient condition for the existence of stochastic stability of the closed-loop system and the consistency controller gain, S h 、S t is an unknown positive definite matrix, G, Z h is the unknown matrix, σ i is a given scalar, γ1 and γ2 are unknown scalars, and other variables are known;

[0126] The sixth inequality (6) is solved using the LMI toolbox in MATLAB. When the given scalar satisfies σ i > 0 (i = 1, ..., N), if there exists a scalar γ1 > 0, γ2 > 0 and a positive definite matrix S h 、S t , and matrices G, Z h , so that the sixth inequality (6) holds true, then the closed-loop system is stochastically stable and satisfies the preset H ∞ performance index, and thereby obtain the consistency controller gain (7), the consistency controller gain (7):

[0127] K(ρ(k))=Z h G -1 (twenty four)

[0128] If there is no feasible solution to inequality (6), the consistency controller gain cannot be obtained.

[0129] In the above example, based on stochastic stability theory, a suitable Lyapunov function was constructed to derive sufficient conditions for stochastic stability of the closed-loop system and the existence of a consistent controller gain. This was then solved using MATLAB's LMI toolbox. This method effectively calculates the consistent controller gain, ensuring stable operation of the closed-loop system in stochastic environments and improving the stability and reliability of multi-agent systems in complex communication environments.

[0130] Preferably, in said S5, a residual evaluation function and a threshold are defined based on the residual evaluation mechanism and the residual signal, and a result of whether the multi-agent system has a fault is obtained by using the residual evaluation function and the threshold, including:

[0131] The residual evaluation function and threshold are defined based on the residual evaluation mechanism and residual signal:

[0132]

[0133] in, It is expressed as the set of all permissible perturbations, sup is the supremum calculation, f i (k) = 0 means that in the absence of a fault, Indicates that the disturbance is within the allowable range;

[0134] Substitute the value of the residual signal into the residual evaluation function to obtain the result of the residual evaluation function, and compare the result of the residual evaluation function with the threshold.

[0135]

[0136] If the result of the residual evaluation function is greater than the threshold, it is determined that a fault has occurred and an alarm message is generated.

[0137] If the result of the residual evaluation function is less than or equal to the threshold, it is determined that no fault has occurred.

[0138] Specifically, based on the adaptive event triggering mechanism (12), the fault detection observer can obtain the residual signal r by formula (2): i (k), and then the residual evaluation function is calculated by formula (8) and formula (9) and threshold Finally, the logic of formula (10) is used to determine whether a fault occurs.

[0139] In the above embodiment, a residual evaluation function and threshold are defined based on the residual evaluation mechanism. System faults are determined by comparing the residual evaluation function result with the threshold, and an alarm is generated promptly when a fault occurs. This approach provides a clear and intuitive basis for fault diagnosis, improving the timeliness and accuracy of fault detection.

[0140] The following describes the processing of a multi-agent system fault detection method affected by channel fading through a specific parameter example. The specific implementation method is as follows:

[0141] The communication topology diagram of a vertical take-off and landing aircraft system is as follows: Figure 3 As shown, the mathematical model is formula (1), where the system parameters are:

[0142]

[0143] Among them, ρ(k)=col{ρ1,ρ2}, ρ1∈[-0.5,0.5], ρ2∈[-2,2].

[0144] Assuming that the order of the Rice fading channel model is L = 2, the mathematical expectation of the channel coefficient is set as: At the same time, the corresponding variance is set to: In addition, it is assumed that the external disturbance input acting on the aircraft can be expressed as:

[0145] ω1(k)=3cos(k), ω2(k)=-2.8sin(1.5k), ω3(k)=2.4cos(1.5k), ω4(k)=-2cos(2.5k),

[0146] ω5(k)=1.8sin(2.5k)

[0147] Assume that the fault occurs on aircraft 2 and 4, and the specific form can be expressed as follows:

[0148]

[0149] By solving the linear matrix inequality (4), we can get the joint H ∞ / H - The performance indicators are γ1 = 0.0204 and β = 0.0760. In addition, the gain of the fault detection observer can be obtained:

[0150]

[0151] According to the threshold expression (9), the threshold of an agent is calculated as: J th1 =1.751, J th2 =1.969, J th3 =1.063, J th4 =1.331, J th5 =0.984. Reference Figure 4 The simulation results from Figure 4 It can be observed that after the fault occurs, the evaluation function values ​​of multi-agents 2 and 4 quickly exceed the threshold. This shows that the fault detection scheme is very effective for linear variable parameter multi-agent system networks.

[0152] For any non-zero initial state, when θ(k) = 0, the closed-loop system is asymptotically stable. When θ(k) ≠ 0, the system consistency is verified by the above method. The specific implementation method is as follows:

[0153] In the simulation, the initial state of the system is set as:

[0154] x1(0)=[0.46 -0.73 0.18 -0.37] T , x2(0)=[-0.87 0.45 -0.51 0.29] T ,

[0155] x3(0)=[0.62 0.54 -0.72 0.56] T , x4(0)=[-0.75 -0.26 0.28 0.45] T ,

[0156] x5(0)=[-0.53 0.74 0.57 -0.35] T ,

[0157] The initial conditions for the trigger parameter threshold are given as:

[0158] σ 1,0 =0.025,σ 2,0 =0.036,σ 3,0 =0.028,σ 4,0 =0.035,σ 5,0 =0.043

[0159] By solving the linear inequality (8), we can obtain the following results:

[0160]

[0161]

[0162] γ2=9.3354,

[0163]

[0164] Based on the above results, according to the control protocol (3), we can get Figure 5 . refer to Figure 5 It can be seen that the consistency goal has been successfully achieved.

[0165] Compared with existing technologies, this invention addresses the channel fading phenomenon in multi-agent systems by introducing a polyhedral linear variable parameter model and using the Rice fading model to describe unreliable wireless network communications between agents. It also provides a design method for a fault detection observer and a consistency controller in a channel fading environment. This method can generate a residual signal that is sensitive to faults and robust to disturbances under the influence of channel fading, effectively reducing network communication pressure while ensuring consistent control performance. It has practical potential and significant advantages.

[0166] Example 2: Figure 7 As shown, another embodiment of the present invention provides a multi-agent system fault detection device affected by channel fading, comprising:

[0167] System state description module, used to describe the multi-agent system in the form of a linear variable parameter system model;

[0168] An observer and controller building module, which is used to build a fault detection observer for detecting faults in a multi-agent system and a consensus controller for unifying the behaviors of each agent in the multi-agent system based on a linear variable parameter system model;

[0169] A gain calculation module is used to calculate the gain of the fault detection observer based on the asymptotic stability theory of the error system and the sufficient conditions for the existence of the gain of the fault detection observer, and to calculate the gain of the consistency controller based on the stochastic stability theory of the closed-loop system and the sufficient conditions for the existence of the gain of the consistency controller;

[0170] a state update module, configured to update the input of the linear variable parameter system model using the gain of the fault detection observer and the gain of the consistency controller, update the output based on the input of the linear variable parameter system model, and perform difference calculation between the updated output of the linear variable parameter system model and the output of the fault detection observer to obtain a residual signal value;

[0171] The fault analysis module is used to perform fault judgment on the residual signal value based on the residual evaluation mechanism, so as to obtain the result of whether the multi-agent system has a fault.

[0172] Preferably, in the system state description module, the state space of the multi-agent system is described in the form of a linear variable parameter system model, including:

[0173] Determine the number of agents in the multi-agent system as N, and the linear variable parameter system model for the i-th agent is expressed as:

[0174]

[0175] in, is the state vector, is the output vector, is the control input, i∈N, For external disturbances, is the fault signal to be detected in the multi-agent system, and the multi-agent system matrix C and D ω and D f are all known constant matrices, ρ(k) is the scheduling vector, A(ρ(k)), B u (ρ(k)),B ω (ρ(k)) and B f (ρ(k)) is a time-varying matrix that depends on the scheduling vector ρ(k). The scheduling parameter ρ(k) satisfies the following conditions:

[0176]

[0177] Among them, ρ s and is a known scalar.

[0178] Example 8: Another embodiment of the present invention provides a multi-agent system fault detection device affected by channel fading, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the multi-agent system fault detection method affected by channel fading as described in any one of the above items.

[0179] Example 4: Another embodiment of the present invention provides a multi-agent system fault detection device affected by channel fading, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the multi-agent system fault detection method affected by channel fading as described above is implemented.

[0180] Embodiment 4: Another embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the multi-agent system fault detection method affected by channel fading as described above is implemented.

[0181] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0182] In the several embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For example, the division of units is merely a logical functional division. In actual implementation, other division methods may be used, such as combining or integrating multiple units or components into another system, or ignoring or not implementing certain features.

[0183] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected based on actual needs to achieve the objectives of the embodiments of the present invention.

[0184] 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 principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for fault detection in a multi-agent system affected by channel fading, characterized in that: The steps include: S1. Describe the multi-agent system in the form of a linear variable parameter system model; S2. Construct a fault detection observer for detecting faults in a multi-agent system based on a linear variable parameter system model and a consistency controller for unifying the behaviors of each agent in the multi-agent system; S3. Calculate the gain of the fault detection observer based on the asymptotic stability theory of the error system and the sufficient conditions for the existence of the gain of the fault detection observer, and calculate the gain of the consistency controller based on the stochastic stability theory of the closed-loop system and the sufficient conditions for the existence of the gain of the consistency controller; S4. updating the input of the linear variable parameter system model using the fault detection observer gain and the consistency controller gain, updating the output based on the input of the linear variable parameter system model, and performing a difference calculation between the updated output of the linear variable parameter system model and the output of the fault detection observer to obtain a residual signal value; S5. Perform fault judgment on the residual signal value based on the residual evaluation mechanism to determine whether the multi-agent system has failed.

2. The multi-agent system fault detection method according to claim 1, characterized in that: In S1, the multi-agent system is described as a linear variable parameter system model, including: Determine the number of agents in the multi-agent system as N, and the linear variable parameter system model for the i-th agent is expressed as: in, is the state vector, is the output vector, is the control input, i∈N, For external disturbances, is the fault signal to be detected in the multi-agent system, and the matrices C and D ω and D f are all known constant matrices, ρ(k) is the scheduling vector, A(ρ(k)), B u (ρ(k)),B ω (ρ(k)) and B f (ρ(k)) is a time-varying matrix that depends on the scheduling vector ρ(k). The scheduling parameter ρ(k) satisfies the following conditions: ρ(k)=col{ρ1(k),ρ2(k),...,ρ nρ (k)} Among them, ρ s and is a known scalar.

3. The multi-agent system fault detection method according to claim 1, characterized in that: In S2, a fault detection observer for detecting faults in a multi-agent system and a consistency controller for unifying the behaviors of agents in the multi-agent system are constructed based on a linear variable parameter system model, including: Assume that the estimated state of the fault detection observer is The output of the fault detection observer is The residual signal is r i (k) Constructing a fault detection observer based on the time-varying matrix and the constant matrix in the linear variable parameter system model, wherein the fault detection observer is: Among them, H a (ρ(k)) and H b (ρ(k)) are both fault observer gains; Assume that the control input of agent i at time k is u i (k) Based on the scheduling vector in the linear variable parameter system model and introducing an event trigger mechanism, a consistency controller is constructed. The consistency controller is: in, It is the observation status information when the event is triggered, that is, the triggering moment When is the current broadcast state of agent i, agent j is the neighbor agent of agent i, is the state information transmitted by the neighboring agent, α0(k) is the initial channel coefficient, α s (k) is the channel fading coefficient, s=0,1,...,l k , K(ρ(k)) is the consistency controller gain.

4. The multi-agent system fault detection method according to claim 1, characterized in that: In S3, the gain of the fault detection observer in the current state is calculated based on the sufficient conditions for the asymptotic stability of the error system and the existence of the gain of the fault detection observer, including: Introducing an error system Where, ε(k) = col{ε1(k),ε2(k),...,ε N (k)},ω(k)=col{ω1(k),ω2(k),...,ω N (k)}, f(k)=col{f1(k),f2(k),…,f N (k)},r(k)=col{r1(k),r2(k),…,r N (k)}, The Lyapunov function is constructed based on the error system. The Lyapunov function is: According to the Lyapunov function, we can deduce: in, When the first inequality holds, the error system is asymptotically stable and satisfies H ∞ Disturbance rejection level, the first inequality is: By applying Schur's complement lemma, we get the second inequality: Multiply both sides of the second inequality by U and U respectively T ,make H a (ρ(k))=X -1 Y ah , H b (ρ(k))=X -1 Y bh , while introducing the condition So that the third inequality holds, the third inequality is: in, The third inequality is a sufficient condition for the asymptotic stability of the error system and the existence of the fault detection observer gain, Q h and Q t is an unknown positive definite matrix, X, Y ah 、Y bh (h, t = 1, 2, ..., λ) is the unknown matrix, γ1 and β are unknown scalars, and the other variables are known; Use the LMI toolbox in MATLAB to solve the third inequality. If there are scalars γ1>0, β>0 and a positive definite matrix Q h , Q t , and matrices X, Y ah 、Y bh (h, t = 1, 2, ..., λ), so that the third inequality holds, and the fault detection observer gain is obtained. The fault detection observer gain is expressed as:

5. The multi-agent system fault detection method according to claim 1, characterized in that: In S3, based on the closed-loop system stochastic stability theory and the sufficient conditions for the existence of the consistency controller gain, the gain of the consistency controller in the current state is calculated, including: Construct a Lyapunov function: V(χ(k))=V1(k)+V2(k), in, Based on the random stability theory, the stability condition expression is derived: The following results are derived: Rewrite q(k) and g(k) from concentrated variable form to component form: q(k)=col{q1(k),q2(k),...,q N (k)}, g(k)=col{q(k-1),q(k-2),...,q(kL)}, in, Based on the closed-loop system in H ∞ It is stochastically stable under the norm constraint and must satisfy the fourth inequality: Using Schur's complement lemma, we can construct the fifth inequality: in, Defining variables Multiply both sides of the fifth inequality by J T With J, we get the sixth inequality and introduce the condition So that the sixth inequality holds, the sixth inequality is: in, Among them, the sixth inequality is a sufficient condition for the stochastic stability of the closed-loop system and the existence of the consistent controller gain, S h 、S t is an unknown positive definite matrix, G, Z h is the unknown matrix, σ i is a given scalar, γ1 and γ2 are unknown scalars, and other variables are known; Using the LMI toolbox in MATLAB to solve the sixth inequality, when the given scalar satisfies σ i > 0 (i = 1, ..., N), if there exists a scalar γ1 > 0, γ2 > 0 and a positive definite matrix S h 、S t , and matrices G, Z h , so that the sixth inequality holds true, and the consistency controller gain is obtained, which is: K(ρ(k))=Z h G -1 。 6. The multi-agent system fault detection method according to any one of claims 1 to 5, characterized in that: In S5, a fault judgment is performed on the residual signal value based on the residual evaluation mechanism, thereby obtaining a result of whether a fault occurs in the multi-agent system, including: The residual evaluation function and threshold are defined based on the residual evaluation mechanism: in, It is expressed as the set of all permissible perturbations, sup is the supremum calculation, f i (k) = 0 means that in the absence of a fault, Indicates that the disturbance is within the allowable range; Substitute the residual signal value into the residual evaluation function to obtain the result of the residual evaluation function, and compare the result of the residual evaluation function with the threshold. If the result of the residual evaluation function is greater than the threshold, it is determined that the multi-agent system has failed and an alarm message is generated. If the result of the residual evaluation function is less than or equal to the threshold, it is determined that the multi-agent system has not failed.

7. A multi-agent system fault detection device affected by channel fading, characterized in that: include: System state description module, used to describe the multi-agent system in the form of a linear variable parameter system model; An observer and controller building module, which is used to build a fault detection observer for detecting faults in a multi-agent system and a consensus controller for unifying the behaviors of each agent in the multi-agent system based on a linear variable parameter system model; A gain calculation module is used to calculate the gain of the fault detection observer based on the asymptotic stability theory of the error system and the sufficient conditions for the existence of the gain of the fault detection observer, and to calculate the gain of the consistency controller based on the stochastic stability theory of the closed-loop system and the sufficient conditions for the existence of the gain of the consistency controller; a state update module, configured to update the input of the linear variable parameter system model using the gain of the fault detection observer and the gain of the consistency controller, update the output based on the input of the linear variable parameter system model, and perform difference calculation between the updated output of the linear variable parameter system model and the output of the fault detection observer to obtain a residual signal value; The fault analysis module is used to perform fault judgment on the residual signal value based on the residual evaluation mechanism, so as to obtain the result of whether the multi-agent system has a fault.

8. The multi-agent system fault detection device according to claim 7, characterized in that: In the system state description module, the multi-agent system is described as a linear variable parameter system model, including: Determine the number of agents in the multi-agent system as N, and the linear variable parameter system model for the i-th agent is expressed as: in, is the state vector, is the output vector, is the control input, i∈N, For external disturbances, is the fault signal to be detected in the multi-agent system, and the matrices C and D ω and D f are all known constant matrices, ρ(k) is the scheduling vector, A(ρ(k)), B u (ρ(k)),B ω (ρ(k)) and B f (ρ(k)) is a time-varying matrix that depends on the scheduling vector ρ(k). The scheduling parameter ρ(k) satisfies the following conditions: ρ(k)=col{ρ1(k),ρ2(k),...,ρ nρ (k)} Among them, ρ s and is a known scalar.

9. A multi-agent system fault detection device affected by channel fading, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the multi-agent system fault detection method affected by channel fading as described in any one of claims 1 to 6 is implemented.

Images