Leaderless unmanned vehicle formation system fault estimation and fault-tolerant method

Abstract

The application provides a leaderless unmanned vehicle formation system fault estimation and fault-tolerant method, adopts an adaptive fault observer to estimate faults of each unmanned vehicle, and then adopts an adaptive fault-tolerant control rate to compensate the estimated faults through fault-tolerant consensus control, wherein the adaptive fault observer can make state estimation errors and fault estimation errors of each unmanned vehicle satisfy a consensus ultimate bounded condition, online real-time monitoring and fault-tolerant control of the leaderless unmanned vehicle formation system can be realized, and robustness and safety of the system are improved.

Description

Technical Field

[0001] This invention belongs to the field of multi-agent control technology, and in particular relates to a fault estimation and fault tolerance method for leaderless unmanned vehicle platooning systems. Background Technology

[0002] An autonomous vehicle platooning system is a unified whole in which multiple autonomous vehicles are organized together and operate collaboratively through information exchange. It resolves contradictions and conflicts among the members during the task process through consistency, avoids the blindness of individual vehicles solving global problems, and can better respond to environmental deviations.

[0003] Fault estimation in autonomous vehicle platooning systems involves monitoring and analyzing the system's state to determine the presence of faults and identify their type and magnitude. Due to the high degree of autonomy and distributed control inherent in platooning systems, a fault in a single vehicle can trigger a chain reaction, leading to unpredictable and abnormal behavior. Fault-tolerant control in platooning systems refers to designing appropriate methods and strategies to ensure that when multiple vehicles experience faults, the faulty vehicle maintains a certain level of performance while minimizing the impact on other functioning vehicles, thus preventing a degrade or interruption of the overall system performance. Therefore, introducing fault estimation and fault-tolerant mechanisms into platooning systems is essential.

[0004] Distributed adaptive observers are algorithms for observing the state of autonomous vehicles. They utilize information from each vehicle to estimate its state and faults. In this algorithm, each autonomous vehicle is adaptive, determining its own observation errors and parameters based on its current state and output. Fault-tolerant algorithms are designed based on the acquired fault information to achieve fault compensation, reducing the adverse effects of faults on the system, enabling self-monitoring, and maintaining the system's normal operation and stability.

[0005] Currently, many fault detection methods for autonomous vehicle platooning systems cannot accurately estimate the size and nature of faults, nor do they address the fault source; they merely detect faults without using the information to compensate for the faulty vehicles. Therefore, it is crucial to quickly detect faults, estimate their size, nature, and type, and utilize this information for fault compensation. From both theoretical and engineering perspectives, fault estimation and fault-tolerant control in autonomous vehicle platooning systems are essential for the safe and stable operation of the system. Summary of the Invention

[0006] To address the aforementioned issues, this invention provides a fault estimation and fault-tolerance method for leaderless unmanned vehicle platooning systems. This method enables online real-time monitoring and fault-tolerant control of leaderless unmanned vehicle platooning systems, thereby improving the robustness and safety of the system.

[0007] A fault estimation and fault tolerance method for a leaderless unmanned vehicle platooning system is proposed. The method uses an adaptive fault observer to estimate the faults of each unmanned vehicle, and then uses an adaptive fault-tolerant control law to compensate for the estimated faults through fault-tolerant consistency control. The adaptive fault observer enables the state estimation error and fault estimation error of each unmanned vehicle to satisfy the consistent eventual bounded condition.

[0008] Furthermore, the method for obtaining the adaptive fault estimator is as follows:

[0009] Step 1: Establish the system model as follows:

[0010]

[0011] Where, x i (t) represents the actual value of the state vector of the i-th autonomous vehicle, u i (t) represents the actual value of the input vector of the i-th autonomous vehicle, y i (t) represents the actual value of the output vector of the i-th autonomous vehicle, f i (t) represents the actual fault value of the i-th unmanned vehicle, which can characterize the size and form of the fault; This represents the actual state vector x i (t) Find the first derivative, which is the rate of change of the actual state vector. Matrices A, B, C, and E are known real matrices.

[0012] Step 2: Design a linear adaptive fault observer based on the system model as follows:

[0013]

[0014] in, It is the estimated state vector value of the i-th autonomous vehicle. It is the estimated output vector value of the i-th autonomous vehicle. R is the fault estimate of the i-th unmanned vehicle, and R is the gain matrix of the undetermined adaptive fault observer. It is the state vector estimate. The rate of change;

[0015] Step 3: Define the state estimation error e of the i-th autonomous vehicle xi (t), Fault estimation error e fi (t) and output estimation error e yi (t) is as follows:

[0016]

[0017] Step 4: Define the dynamic error equation as follows:

[0018]

[0019] in, Let e ​​be the state estimation error of the i-th autonomous vehicle. xi The rate of change of (t);

[0020] The state estimation error of all autonomous vehicles e xi (t), Fault estimation error e fi (t), Output estimation error e yi (t) and the actual fault value f i (t) forms the global variables as follows:

[0021]

[0022] Where i = 1, 2, ..., N, N represents the number of autonomous vehicles in the platoon, e x (t) is a global variable concerning the state estimation error, e f (t) is a global variable concerning the fault estimation error, e y f(t) is a global variable relating to the output estimation error, and f(t) is a global variable relating to the actual fault value. For the fault estimate;

[0023] The dynamic error equation can be expressed in global variable form as follows:

[0024]

[0025] Among them, I N It is a unit vector. Indicates the Kronecker product. For global variable e x The rate of change of (t);

[0026] Obtain the fault estimation error e f The derivative of (t) with respect to time is as follows:

[0027]

[0028] in, For global variable e f The rate of change of (t), global variable rate of change, The rate of change of the global variable f(t);

[0029] Step 5: Design the adaptive fault observer as follows:

[0030]

[0031] Where Θ=Θ T >0 represents the set adaptive learning rate. For global variable e y The rate of change of (t), where θ1 is the rate of change. The weight matrix is ​​set, and θ2 is the global variable e. y The weight matrix of (t).

[0032] Furthermore, to ensure that the state estimation errors and fault estimation errors of each autonomous vehicle satisfy the consistent eventual bounded condition, the values ​​of the real matrices A, B, C, E, the gain matrix R, and the weight matrices θ1 and θ2 of the adaptive fault observer must satisfy the following conditions:

[0033] There exist a symmetric positive definite matrix P and a constant ε > 0 such that:

[0034]

[0035]

[0036] Where I is the identity matrix, Π1, Π2, Π4, and Y are auxiliary matrices, and we have:

[0037]

[0038] By R=P -1 Y obtains the gain matrix R of the adaptive fault observer, and then calculates the real matrix A, B, C, E and the weight matrix θ1, θ2 using formulas (9) and (10).

[0039] Furthermore, the adaptive fault-tolerant control rate u(t) is as follows:

[0040]

[0041] Where Γ is the feedback gain matrix, B * Is it satisfied (I) N -BB * The matrix E = 0, where ξ(t) is the global relative state error. e (t) represents the global relative tracking error, and we have:

[0042]

[0043] in, Let be the relative state error of the 1st to Nth autonomous vehicles in the platoon. Let be the relative tracking error of the 1st to Nth autonomous vehicles in the platoon, and we have:

[0044]

[0045] Where, ξ i (t) represents the relative state error of the i-th autonomous vehicle in the autonomous vehicle platoon, x j (t) represents the actual value of the state vector of the j-th unmanned vehicle, a ij This represents the connection relationship between driverless car i and driverless car j, when a ij =1 indicates that the two driverless vehicles can exchange information. ei (t) represents the relative tracking error of the i-th autonomous vehicle in the autonomous vehicle platoon, e xj (t) represents the state estimation error of the j-th unmanned vehicle.

[0046] Furthermore, when the adaptive fault-tolerant control rate achieves fault-tolerant consistency control, the global relative state error ξ(t) must satisfy the following condition:

[0047] There exist a symmetric positive definite matrix P and a dimension matrix W, with constants β1 > 0 and β2 > 0, such that:

[0048] Γ=-B T P (18)

[0049] E = -BW (19)

[0050] β1||W||+β2||Γ||≤0 (20)

[0051]

[0052] in, It is a Laplace matrix.

[0053] Furthermore, for all autonomous vehicles, if their position and consistency position error tend to remain unchanged, it indicates that the autonomous vehicle formation has achieved consistency, and the relative positions between each autonomous vehicle remain unchanged. At this time, if any autonomous vehicle experiences a constant or random fault, the adaptive fault-tolerant control rate will automatically start to compensate for the faulty autonomous vehicle, so that the faulty autonomous vehicle can regain consistency. The position and consistency position error of the other autonomous vehicles that deviated at the moment the fault was added will also regain consistency. After that, the relative positions between each autonomous vehicle remain unchanged and continue to operate stably.

[0054] Beneficial effects:

[0055] 1. This invention provides a fault estimation and fault tolerance method for a leaderless unmanned vehicle platooning system. It employs an adaptive fault observer to estimate the faults of each unmanned vehicle, and then uses an adaptive fault-tolerant control law to compensate for the estimated faults through fault-tolerant consistency control. The adaptive fault observer enables the state estimation error and fault estimation error of each unmanned vehicle to satisfy a consistent eventual bounded condition, thereby realizing online real-time monitoring and fault-tolerant control of the leaderless unmanned vehicle platooning system and improving the robustness and safety of the system.

[0056] 2. This invention provides a fault estimation and fault tolerance method for a leaderless unmanned vehicle platooning system. By monitoring and diagnosing each unmanned vehicle, an observer is used to estimate the system state and faults. An adaptive algorithm is used to update the observer and controller. Parameters can be adjusted according to the real-time state to achieve adaptive control, thereby ensuring the reliability and robustness of the entire system.

[0057] 3. This invention provides a fault estimation and fault tolerance method for leaderless unmanned vehicle platooning systems. It gives sufficient conditions for the stability of the global dynamic error equation in the form of matrix inequalities, and reduces the number of constraints by rearranging the inequalities. Finally, based on fault estimation, an active fault-tolerant control algorithm is designed to compensate for the fault, proving the stability of the global relative tracking error system and making up for the difficulty in meeting the fault estimation performance requirements in traditional adaptive fault estimation observers. Attached Figure Description

[0058] Figure 1 Flowchart of a fault estimation method for a leaderless unmanned vehicle platooning system provided as an embodiment of the present invention;

[0059] Figure 2 For communication topology;

[0060] Figure 3 Given a constant fault, state x and its observed values;

[0061] Figure 4 Output y and its observed values ​​under constant fault conditions;

[0062] Figure 5 For constant fault conditions, the fault f and its observed values ​​are given.

[0063] Figure 6 This represents the estimation error of the output y under constant fault conditions.

[0064] Figure 7 Let x be the state under random faults and its observed values;

[0065] Figure 8 Output y and its observed values ​​under random fault conditions;

[0066] Figure 9For a random fault, consider the fault f and its observed values.

[0067] Figure 10 The estimation error of output y under random fault conditions;

[0068] Figure 11 The vehicle position under constant fault conditions;

[0069] Figure 12 This refers to the consistency position error under constant fault conditions.

[0070] Figure 13 The vehicle's location under random fault conditions;

[0071] Figure 14 This represents the consistency position error under random fault conditions. Detailed Implementation

[0072] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0073] This invention addresses leaderless autonomous vehicle platooning systems by designing a novel distributed adaptive fault estimation observer using a fast adaptive fault estimation algorithm for fault detection. Finally, adaptive fault-tolerant control is implemented based on the observer, belonging to the field of fault estimation and fault tolerance technology for multi-agent systems.

[0074] Specifically, a fault estimation and fault tolerance method for a leaderless unmanned vehicle platooning system is proposed. An adaptive fault observer is used to estimate the faults of each unmanned vehicle, and then an adaptive fault-tolerant control rate is used to compensate for the estimated faults through fault-tolerant consistency control. The adaptive fault observer enables the state estimation error and fault estimation error of each unmanned vehicle to satisfy the consistent eventual bounded condition.

[0075] For all autonomous vehicles, if their position and consistency position error tend to remain unchanged, it means that the autonomous vehicle formation has achieved consistency and the relative positions between each autonomous vehicle remain unchanged. At this time, if any autonomous vehicle fails, the adaptive fault-tolerant control rate will automatically start to compensate for the failure and restore the consistency of the failure autonomous vehicle. The position and consistency position error of the other autonomous vehicles that deviated at the moment the failure was added will also return to consistency. After that, the relative positions between each autonomous vehicle remain unchanged and continue to operate stably.

[0076] Furthermore, consider a leaderless linear platooning system of N autonomous vehicles. To obtain fault estimation information, without loss of generality, assume that a fault observer is established for the i-th autonomous vehicle; where, as... Figure 1 As shown, the method for obtaining the adaptive fault estimator includes the following steps:

[0077] Step 1: Establish the system model as follows:

[0078]

[0079] Where, x i (t)∈R n Let u represent the actual value of the state vector of the i-th autonomous vehicle. i (t)∈R g Let y represent the actual value of the input vector of the i-th autonomous vehicle. i (t)∈R p f represents the actual value of the output vector of the i-th autonomous vehicle. i (t)∈R q This represents the actual fault value of the i-th unmanned vehicle, which can characterize the magnitude and form of the fault; This represents the actual state vector x i (t) Find the first derivative, which is the rate of change of the actual state vector. Matrices A, B, C, and E are known real matrices.

[0080] Step 2: Design a linear adaptive fault observer based on the system model as follows:

[0081]

[0082] in, It is the estimated state vector value of the i-th autonomous vehicle. It is the estimated output vector value of the i-th autonomous vehicle. R is the fault estimate of the i-th unmanned vehicle, and R is the gain matrix of the undetermined adaptive fault observer. It is the state vector estimate. The rate of change;

[0083] Step 3: Define the state estimation error e of the i-th autonomous vehicle xi (t), Fault estimation error e fi (t) and output estimation error e yi (t) is as follows:

[0084]

[0085] Step 4: Combining formula (2), the dynamic error equation is defined as follows:

[0086]

[0087] in, Let e ​​be the state estimation error of the i-th autonomous vehicle. xi The rate of change of (t);

[0088] The state estimation error of all autonomous vehicles e xi (t), Fault estimation error e fi (t), Output estimation error e yi (t) and the actual fault value f i (t) forms the global variables as follows:

[0089]

[0090] Where i = 1, 2, ..., N, N represents the number of autonomous vehicles in the platoon, e x (t) is a global variable concerning the state estimation error, e f (t) is a global variable concerning the fault estimation error, e y f(t) is a global variable relating to the output estimation error, and f(t) is a global variable relating to the actual fault value. For the fault estimate, T is a global variable, and T denotes transpose.

[0091] The dynamic error equation can be expressed in global variable form as follows:

[0092]

[0093] Among them, I N It is a unit vector. Indicates the Kronecker product. For global variable e x The rate of change of (t);

[0094] Obtain the fault estimation error e f The derivative of (t) with respect to time is as follows:

[0095]

[0096] in, For global variable e f The rate of change of (t), global variable rate of change, The rate of change of the global variable f(t);

[0097] Step 5: Design the adaptive fault observer as follows:

[0098]

[0099] Where Θ=Θ T >0 represents the set adaptive learning rate. For global variable e y The rate of change of (t), where θ1 is the rate of change. The weight matrix is ​​set, and θ2 is the global variable e.y The weight matrix of (t).

[0100] Step 6: The following fast adaptive fault estimation algorithm can minimize the state estimation error e. x (t) and fault estimation error e f (t) Uniformly final bounded condition;

[0101] Specifically, to ensure that the state estimation errors and constant fault estimation errors of each autonomous vehicle satisfy the consistent eventual bounded condition, the values ​​of the real matrices A, B, C, E, the gain matrix R, and the weight matrices θ1 and θ2 of the adaptive fault observer must satisfy the following conditions:

[0102] There exists a symmetric positive definite matrix P∈R n×n And the constant ε > 0, such that:

[0103]

[0104]

[0105] Where I is the identity matrix, Π1, Π2, Π4, and Y are auxiliary matrices, and we have:

[0106]

[0107] Based on the above, the global variable e of the state estimation error can be obtained. x (t) and the global variable e of fault estimation error f (t) is uniformly bounded, thus the fault estimation observer is stable under the fast adaptive fault estimation algorithm of the adaptive fault observer designed by formula (8), and can effectively track the fault value. Through R = P -1 Y obtains the gain matrix R of the adaptive fault observer, and then calculates the real matrix A, B, C, E and the weight matrix θ1, θ2 using formulas (9) and (10).

[0108] Furthermore, the derivation process of the adaptive fault-tolerant control rate u(t) is as follows:

[0109] Define the adaptive fault-tolerant control rate u of the i-th autonomous vehicle. i (t) is as follows:

[0110]

[0111] Where Γ is the feedback gain matrix, B * Is it satisfied (I) N -BB * The matrix E = 0, where ξ(t) is the global relative state error. e (t) represents the global relative tracking error, a ijThis represents the connection relationship between driverless car i and driverless car j, when a ij =1 indicates that the two driverless vehicles can exchange information. It is the estimated state vector value of the i-th autonomous vehicle. It is the estimated state vector value of the j-th autonomous vehicle. It is the fault estimate of the i-th unmanned vehicle.

[0112] The relative tracking error is defined as follows:

[0113]

[0114] Where, ξ i (t) represents the relative state error of the i-th autonomous vehicle in the autonomous vehicle platoon, x j (t) represents the actual value of the state vector of the j-th unmanned vehicle, ξ ei (t) represents the relative tracking error of the i-th autonomous vehicle in the autonomous vehicle platoon, e xj (t) represents the state estimation error of the j-th unmanned vehicle;

[0115] Define global variables as follows:

[0116]

[0117] in, Let be the relative state error of the 1st to Nth autonomous vehicles in the platoon. ξ(t) represents the relative tracking error of the 1st to Nth unmanned vehicles in the platoon; ξ(t) represents the global relative state error. e (t) represents the global relative tracking error;

[0118] The relative tracking error of equation (13) can then be written in the following global form:

[0119]

[0120] in, It is a Laplace matrix;

[0121] Fault tolerance control ratio can be written in global form:

[0122]

[0123] The global relative tracking error ξ(t) can be simplified to the following form:

[0124]

[0125] Furthermore, this invention can also provide a condition for the global relative state error ξ(t) to be asymptotically stable. Specifically, when the adaptive fault-tolerant control law achieves fault-tolerant consistency control, the global relative state error ξ(t) needs to satisfy the following condition:

[0126] There exist a symmetric positive definite matrix P and a dimension matrix W, with constants β1 > 0 and β2 > 0, such that:

[0127] Γ=-B T P (18)

[0128] E = -BW (19)

[0129] β1||W||+β2||Γ||≤0 (20)

[0130]

[0131] in, The Laplace matrix is ​​used to achieve fault-tolerant consistency control under this condition, and the required parameters in formula (12) are calculated by formulas (18)-(21).

[0132] The following describes in further detail a fault estimation and fault tolerance method for a leaderless unmanned vehicle platooning system provided by the present invention, using specific embodiments.

[0133] Example 1

[0134] This embodiment consists of six unmanned vehicles with identical characteristics, used to simulate the cooperation and interaction process in an unmanned vehicle platooning system. To simulate fault occurrence, a sudden event is introduced, causing one vehicle to experience a constant fault. The communication topology is as follows: Figure 2 As shown.

[0135] The system adopts an undirected graph topology, consisting of... Figure 2 Find the Laplace matrix of the autonomous vehicle platooning system Assume that driverless car 2 experiences a constant fault at the 3rd second, while the other vehicles do not experience faults. The fault description is as follows:

[0136]

[0137] Figure 3 , Figure 4The curves showing the changes of state x and its observed values, and output y and its observed values ​​over time under a constant fault are presented. As can be seen from the figure, under the novel distributed adaptive fault estimation observer, when there is no fault before t=3s and the system is stable, the observed values ​​of state x and output y can track the actual values ​​well through the observer's data processing, indicating that the method can accurately capture and estimate the system state. When a constant fault is introduced at t=3s, the actual values ​​of state x and output y shift downwards significantly, and the observed values ​​almost perfectly match the downward shift of the actual values, indicating that the observer can accurately estimate the system state.

[0138] Figure 5 , Figure 6 The curves showing the time-varying effects of the estimation errors of fault f, its observed values, and output y under constant fault conditions are presented. As can be seen from the figure, under the novel distributed adaptive fault estimation observer, when there is no fault before t=3s, the estimation errors of fault f, its observed values, and output y are equal to 0. When a constant fault is introduced at t=3s, the observer can accurately estimate the parameters of the constant fault. The output estimation error fluctuates momentarily upon the introduction of the constant fault, but then approaches zero. This demonstrates the significant fault estimation performance of the novel distributed adaptive fault estimation observer.

[0139] The simulation results in MATLAB show that the maximum estimation error of the fault is 0.0025, and the maximum estimation errors of the output are 0.1872 and 0.0500, respectively.

[0140] Example 2

[0141] The system state and topology in this embodiment are the same as in Embodiment 1, causing one vehicle to experience a random malfunction. Assume that driverless car 2 experiences a random malfunction at the 3rd second, while the other vehicles remain functioning normally. The malfunction description is as follows:

[0142]

[0143] Figure 7 , Figure 8 The curves showing the changes of state x and its observed values, and output y and its observed values ​​over time under random fault conditions are presented. As can be seen from the figure, under the novel distributed adaptive fault estimation observer, when there is no fault before t=3s and the system is stable, the observed values ​​of state x and output y can accurately track the actual values ​​through the observer's data processing, indicating that the method can accurately capture and estimate the system state. When a random fault is introduced at t=3s, state x and output y and their observed values ​​undergo random fluctuations, but the observed values ​​still track the system state well.

[0144] Figure 9 , Figure 10The curves showing the variation of the estimation error of fault f, its observed value, and output y over time under random fault conditions are presented. As can be seen from the figure, under the novel distributed adaptive fault estimation observer, there is no fault before t=3s, and the estimation error of fault f, its observed value, and output y is equal to 0. When a random fault is introduced at t=3s, the observer output does not lag significantly relative to the actual state value, and the observer can accurately estimate the parameters of the random fault. The output estimation error exhibits a random value within a very small range. This demonstrates that the fault estimation effect of the novel distributed adaptive fault estimation observer is significant.

[0145] MATLAB simulation results show that the maximum estimation error of the fault is 0.0976, and the maximum estimation errors of the output are 0.3133 and 0.1265, respectively.

[0146] Example 3

[0147] To verify the effectiveness and stability of the fault-tolerant control method based on a distributed adaptive observer in the event of a fault in an autonomous vehicle system, this example uses the designed adaptive fault-tolerant control law for numerical simulation. Consider a leaderless autonomous vehicle system with six vehicles. Assume that the second vehicle experiences a constant fault, and the adaptive fault-tolerant controller immediately compensates for the impact of the faulty vehicle after the fault occurs. The constant fault is described as follows:

[0148]

[0149] Figure 11 , Figure 12 The curves showing the changes in the position and consistency position error of the autonomous vehicles under a constant fault over time are presented. As can be seen from the figure, for all autonomous vehicles, the position and consistency position error tend to remain constant at t = 1.95s, indicating that the autonomous vehicle system has achieved consistency, and the relative positions between the vehicles remain unchanged. When a constant fault is introduced at t = 3s, the adaptive fault-tolerant control algorithm is activated. Faulty vehicle 2 regains consistency after t = 3.65s. The position and consistency position errors of the remaining vehicles deviate slightly at the moment the constant fault is introduced, and they regain consistency again at t = 3.45s. Afterwards, the relative positions between the vehicles remain unchanged, and the system continues to operate stably.

[0150] Example 4

[0151] This example considers a leaderless autonomous vehicle system with six vehicles. It assumes that the second vehicle experiences a random failure, and the adaptive fault-tolerant controller immediately compensates for the impact of the failed vehicle. The random failure is described as follows:

[0152]

[0153] Figure 13 , Figure 14The curves showing the changes in vehicle position and consistent position error over time under random faults are presented. As can be seen from the figure, for all autonomous vehicles, the position and consistent position errors tend to remain constant at t = 1.95s, indicating that the autonomous vehicle system has achieved consistency, and the relative positions between the vehicles remain unchanged. When a random fault is introduced at t = 3s, the adaptive fault-tolerant control algorithm is activated. Faulty vehicle 2 regains consistency after t = 3.85s. The position and consistent position errors of the remaining vehicles deviate slightly at the moment the random fault is introduced, and they regain consistency again at t = 3.35s. Afterward, the relative positions between the vehicles remain unchanged, and the system continues to operate stably.

[0154] In summary, this invention addresses leaderless unmanned vehicle platooning systems that require precise understanding of the magnitude and nature of faults while ensuring the faulty vehicles still participate in the overall task. By applying a fast adaptive fault estimation algorithm, a novel distributed adaptive fault estimation observer and fault-tolerant control algorithm are designed. This method monitors and diagnoses each unmanned vehicle, uses the observer to estimate the system state and faults, and updates the observer and controller using an adaptive algorithm. Parameters can be adjusted based on real-time conditions to achieve adaptive control, thereby ensuring the reliability and robustness of the entire system.

[0155] Of course, the present invention may have other various embodiments. Without departing from the spirit and essence of the present invention, those skilled in the art can make various corresponding changes and modifications according to the present invention, but these corresponding changes and modifications should all fall within the protection scope of the appended claims.

Claims

1. A fault estimation and fault tolerance method for a leaderless unmanned vehicle platooning system, characterized in that, An adaptive fault observer is used to estimate the faults of each unmanned vehicle, and then an adaptive fault-tolerant control rate is used to compensate for the estimated faults through fault-tolerant consistency control. The adaptive fault observer enables the state estimation error and fault estimation error of each unmanned vehicle to satisfy the consistent final bounded condition. The method for obtaining the adaptive fault estimator is as follows: Step 1: Establish the system model as follows: （1） in, This represents the actual value of the state vector of the i-th autonomous vehicle. Let represent the actual value of the input vector for the i-th autonomous vehicle. This represents the actual value of the output vector of the i-th autonomous vehicle. This represents the actual fault value of the i-th unmanned vehicle, which can characterize the magnitude and form of the fault; Represents the actual state vector Find the first derivative, which is the rate of change of the actual state vector, and the matrix. Given a real matrix; Step 2: Design a linear adaptive fault observer based on the system model as follows: （2） in, It is the estimated state vector value of the i-th autonomous vehicle. It is the estimated output vector value of the i-th autonomous vehicle. It is the fault estimate of the i-th autonomous vehicle. It is the gain matrix of the undetermined adaptive fault observer; It is the state vector estimate. The rate of change; Step 3: Define the state estimation error of the i-th autonomous vehicle Fault estimation error and output estimation error as follows: （3） Step 4: Define the dynamic error equation as follows: （4） in, The state estimation error of the i-th autonomous vehicle The rate of change; All autonomous vehicle state estimation errors Fault estimation error Output estimation error and actual fault values The global variables are composed as follows: （5） in, N represents the number of autonomous vehicles in the platoon. For the state estimation error, (This is a global variable) For the fault estimation error, Let be a global variable relating to the output estimation error. For the actual value of the fault, For global variables relating to fault estimates; The dynamic error equation can be expressed in global variable form as follows: （6） in, It is a unit vector. Indicates the Kronecker product. global variable The rate of change; Obtaining fault estimation error The derivative with respect to time is as follows: （7） in, global variable rate of change, global variable rate of change, global variable The rate of change; Step 5: Design the adaptive fault observer as follows: （8） in, The set adaptive learning rate, global variable rate of change, rate of change The weight matrix is ​​set. global variable The weight matrix is ​​set.

2. The fault estimation and fault tolerance method for a leaderless unmanned vehicle platooning system as described in claim 1, characterized in that, The real matrix of the adaptive fault observer is such that the state estimation errors and fault estimation errors of each unmanned vehicle satisfy a consistent eventual bounded condition. Gain matrix Weight matrix , The value of needs to satisfy the following conditions: There exists a symmetric positive definite matrix and constant , so that: （9） （10） in, It is the identity matrix. , , Y is an auxiliary matrix, and we have: （11） pass Obtain the gain matrix of the adaptive fault observer Then, the real matrix is ​​calculated using formulas (9) and (10). Weight matrix , .

3. The fault estimation and fault tolerance method for a leaderless unmanned vehicle platooning system as described in claim 1, characterized in that, The adaptive fault-tolerant control rate as follows: （16） in, For the feedback gain matrix, It is to satisfy The matrix, This represents the global relative state error. Let be the global relative tracking error, and we have: （14） in, ~ Let be the relative state error of the 1st to Nth autonomous vehicles in the platoon. ~ Let be the relative tracking error of the 1st to Nth autonomous vehicles in the platoon, and we have: （13） in, The first in the autonomous vehicle platoon The relative state error of an autonomous vehicle For the first The actual value of the state vector of an autonomous vehicle. Indicates driverless car and driverless cars The connection relationship, when This indicates that the two driverless vehicles can exchange information with each other. The first in the autonomous vehicle platoon The relative tracking error of an autonomous vehicle For the first The state estimation error of the autonomous vehicle.

4. The fault estimation and fault tolerance method for a leaderless unmanned vehicle platooning system as described in claim 3, characterized in that, When adaptive fault-tolerant control achieves fault-tolerant consistency control, the global relative state error The conditions that need to be met are: There exists a symmetric positive definite matrix and dimension matrix ,constant , so that: （18） （19） （20） （21） in, It is a Laplace matrix.

5. A fault estimation and fault-tolerance method for a leaderless unmanned vehicle platooning system as described in any one of claims 1 to 4, characterized in that, For all autonomous vehicles, if their position and consistency position error tend to remain constant, it means that the autonomous vehicle formation has achieved consistency and the relative positions between each autonomous vehicle remain unchanged. At this time, if any autonomous vehicle experiences a constant or random fault, the adaptive fault-tolerant control rate will automatically start to compensate for the faulty autonomous vehicle, so that the faulty autonomous vehicle can regain consistency. The position and consistency position error of the other autonomous vehicles that deviated at the moment the fault was added will also regain consistency. After that, the relative positions between each autonomous vehicle remain unchanged and continue to operate stably.

Images