Observer-based multi-vehicle system fault detection method, storage medium and equipment

Abstract

The invention discloses a multi-vehicle system fault detection method based on a distributed observer, a storage medium and equipment, and the method comprises the steps: firstly building a multi-vehicle system model, designing an exponential distributed observer based on information and relative measurement, setting an observation gain and a control gain through the Lyapunov stability theory, and carrying out the fault detection of a multi-vehicle system; the method comprises the following steps: determining a criterion that the output of a multi-vehicle system is consistent, obtaining a relative measurement value of adjacent vehicles through computer analogue simulation, finally setting a vehicle alarm threshold value based on the adjacent measurement value, and when the vehicle runs, if the adjacent measurement value of the vehicle is greater than the set alarm threshold value, effectively alarming the vehicle in real time by a designed fault detection mechanism. According to the invention, based on the fault detection design of the observer, the occupied space of the instrument is reduced, the information resource is fully utilized, and the real-time fault detection of the vehicle is effectively carried out while the vehicle consistency is realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of network security, and particularly relates to the technology of vehicle system fault diagnosis. Specifically, it mainly relates to a multi-vehicle system fault detection method, storage medium and device based on a distributed observer. Background Art

[0002] Due to the rapid development of network communication technology, the consensus of intelligent vehicle systems has been widely applied. For example: formation control of multi-vehicle systems, consensus of underwater autonomous vehicle systems. As is well known, in practical applications, there are differences in the dynamic characteristics of each vehicle system. Therefore, the consensus of heterogeneous vehicle systems has been widely studied. On the other hand, based on the complex communication environment in practical applications, it is necessary to supervise the safety of vehicles. For example, a denial-of-service attack on the vehicle network causes network congestion, resulting in the failure of the vehicle to perform tasks, causing great losses in terms of manpower, economy, etc. Therefore, studying the consensus of heterogeneous vehicle systems and simultaneously performing real-time supervision and diagnosis of vehicles is a necessary research project.

[0003] In the network system, due to the existence of noise, characteristics such as information congestion and inaccurate transmission occur. To improve the accuracy of vehicle system fault detection, observers are introduced to improve the problem of poor performance of the system due to the existence of noise. Observers are mainly divided into centralized observers and distributed observers. Since centralized observers have disadvantages such as complex processing tasks and high risks, while distributed observers have characteristics such as scalability, strong coordination, and excellent convergence, distributed observers are widely used in fault detection systems. On the other hand, some existing safety detectors for vehicle systems occupy a large space. For example, in a micro intelligent vehicle system with a small space and limited information resource interaction, the separate design of the detector and the observer has high requirements for space and information resources, making the detector unable to be applied to the vehicle system. Therefore, studying the fault detection mechanism based on a distributed observer is a meaningful topic.

[0004] In addition to noise, time delay is also inevitable in network communication. The existence of time delay has a great impact on the stability of the vehicle system. However, in the design of traditional multi-vehicle fault detection systems, the interaction of vehicle information either ignores the factor of time delay or the given method has a certain conservatism for the upper bound of time delay. For example, it is difficult to calculate the upper bound of time delay for the stability criterion of the time-delay system obtained by the linear matrix inequality method, which causes great inconvenience in practical applications. In addition to the linear matrix inequality method, the Hanaly inequality and the Razumikhin theorem are also used to deal with time-delay systems. However, the stability criterion of the time-delay system obtained by the Hanaly inequality method has a low conservatism, and the conditions of the Razumikhin theorem are difficult to verify in application. Because the above methods all limit their applications in the actual environment, it is a necessary topic to explore a stability criterion for multi-vehicle systems that is easy to verify and has a low conservatism of time-delay upper bound under time-delay communication, and then accurately implement fault supervision. Summary of the Invention

[0005] In view of the problems in the prior art that 1) when the vehicle system achieves consensus, the separate design of the fault detection device and the observer takes up a large space, and 2) the upper bound of the communication time delay of the multi-vehicle fault detection system has a high conservatism, the present invention proposes a multi-vehicle system fault detection method, storage medium and device based on a distributed observer. First, a multi-vehicle system model is established, an exponential distributed observer based on information and relative measurement is designed, and then, with the help of the Lyapunov stability theory, the observation gain and the control gain are set to ensure the output consensus of the multi-vehicle system. The relative measurement values of adjacent vehicles are obtained through computer simulation. Finally, based on the adjacent measurement values, the vehicle alarm threshold is set. In practical applications, when the vehicle is running, if the adjacent measurement value of the vehicle is greater than the set alarm threshold, the designed fault detection mechanism can effectively alarm the vehicle in real time. The present invention is a fault detection design based on a state observer, which reduces the space occupied by instruments, makes full use of information resources, has a low conservatism for the upper bound of time delay, and effectively detects vehicle faults in real time while achieving vehicle consensus.

[0006] To achieve the above object, the technical solution adopted by the present invention is: a multi-vehicle system fault detection method based on an observer, including the following steps:

[0007] S1. Establish a vehicle system model: Mathematically model the leader and followers of the vehicle, establish the dynamic equation of the vehicle system, generate the parameter matrix of the system, and establish the network communication topology from the leader to the followers.

[0008] S2. Design an exponential observer: Design a distributed observer for the leader according to the parameter matrix obtained in step S1.

[0009] S3. Computer simulation: According to the dynamic equations of the vehicle system in step S1 and the observer in step S2, calculate the observation error and the observation error system; based on the Lyapunov stability theory and the small gain theorem, set the observation gain and the control gain, determine the criterion for the output consistency of the multi-vehicle system, and obtain the relative measurement values of adjacent vehicles; the specific criterion is: if there exist a time delay h and a positive definite matrix P satisfying the following formula:

[0010]

[0011] then there exists a control gain matrix such that the dynamic equations of the vehicle system and the follower dynamic equations achieve output consistency; where, is the identity matrix, d is a real number satisfying d ∈ [-1, 0), S is the leader system matrix, the matrix S is normal, I n is the identity matrix, μ(S) is the largest positive real part of the leader matrix S, and e represents the Euler number;

[0012] S4. Alarm detection: Set a threshold for the relative measurement values of adjacent vehicles obtained in step S3. During the actual operation of the vehicle, when the relative measurement value is greater than the threshold, an alarm is given to determine that there is a fault in the vehicle system and achieve fault detection.

[0013] As an improvement of the present invention, in the step S1, the dynamic equation of the leader of the vehicle system is specifically:

[0014]

[0015] y 0 (t) = Rv(t)

[0016] where represents the state of the leader of the vehicle system, y 0 (t) represents the output value of the leader, S is the leader system matrix, and R is the leader output matrix;

[0017] The dynamic equation of the follower is specifically:

[0018]

[0019] y k (t) = C k x k (t), k = 1,..., N

[0020] where represents the state of the k-th vehicle follower, y k (t) represents the output value of the follower, A k , B k , C k Ek is the parameter matrix of the follower system.

[0021] As another improvement of the present invention, the distributed observer in the step S2 is specifically:

[0022]

[0023] where χ k (t) represents the state estimation value of the k-th vehicle for the leader at time t, v(t - h) represents the state of the leader at time t - h, y k (t), y j (t) respectively represent the output values of the k-th and j-th vehicles, h represents the communication time delay from the leader to the follower, γ is a parameter to be designed satisfying γ > 0, L is the observation gain matrix, S is the leader system matrix. If there is a signal connection between vehicle k and vehicle j, then h kj > 0, otherwise h kj = 0.

[0024] As another improvement of the present invention, during the computer simulation process of the step S3, it is satisfied that: the network topology graph of the leader and the followers is undirected and connected, and there is an edge connecting the leader and all the followers; and there exist matrices Π k and U k satisfying

[0025] Π k S = A k Π k + B k U k + E k

[0026] 0 = C k Π k - R, k = 1, ··· N

[0027] where A k , B k , C k , E k , k = 1, …, N are the parameter matrices of the follower system, S is the leader system matrix, and R is the leader output matrix.

[0028] As yet another improvement of the present invention, the follower system in the step S3 includes a control protocol:

[0029]

[0030] where u k (t) represents the control input of the k-th vehicle, is the forward gain of the k-th vehicle, and K xk is the feedback gain of the k-th vehicle.

[0031] As a further improvement of the present invention, in step S1, the vehicles among the followers are classified into groups of three in sequence and numbered, namely vehicle 1, vehicle 2, and vehicle 3. When an alarm occurs in step S4, the specific fault determination rules are as follows:

[0032] If vehicle 1 and vehicle 2 alarm and vehicle 3 does not alarm, it indicates that vehicle 1 has a fault;

[0033] If vehicle 2 and vehicle 3 alarm and vehicle 1 does not alarm, it indicates that vehicle 3 has a fault;

[0034] If vehicle 1, vehicle 2 alarm and vehicle 3 alarms, it indicates that vehicle 2 has a fault or vehicle 1, vehicle 2, and vehicle 3 all have faults.

[0035] To achieve the above object, the technical solution adopted by the present invention is also: a computer-readable storage medium, on which a computer program is stored, and the computer program is executed by a processor to implement the observer-based multi-vehicle system fault detection method as described in any one of claims 1-6.

[0036] To achieve the above object, the technical solution adopted by the present invention is also: a computer device, including:

[0037] A memory for storing instructions;

[0038] A processor for executing the instructions, so that the computer device performs the operations of the observer-based multi-vehicle system fault detection method as described in any one of claims 1-6.

[0039] Compared with the prior art, the present invention has the following beneficial effects:

[0040] 1. Compared with the fault detection mechanism in which the detector and the observer are designed separately, the observer-based fault detection mechanism designed by the present invention reduces the occupation of instrument space, makes full use of information resources, and realizes real-time fault detection of the vehicle system.

[0041] 2. Compared with the existing consistency criterion, the criterion given in step S3 is easy to verify and has a lower conservatism in the time-delay upper bound. When the leader system matrix is semi-stable (μ(S) ≤ 0), the time delay h can be selected as an arbitrarily large value.

[0042] 3. While the method of the present invention realizes the consistency of the vehicle system, it also realizes real-time supervision of the vehicle, ensures the safety of the vehicle system, and provides a new idea for the research of multi-vehicle system fault detection. Description of the Drawings

[0043] Figure 1It is the flow chart of the steps of the method of the present invention;

[0044] Figure 2 It is the structure diagram of observer-based fault detection in Embodiment 1 of the present invention;

[0045] Figure 3 It is the structure diagram of the distributed observer of the vehicle system in Embodiment 1 of the present invention;

[0046] Figure 4 It is the communication topology diagram of the vehicle system in Embodiment 1 of the present invention;

[0047] Figure 5 It is the coordination diagram of the vehicle system in Embodiment 1 of the present invention;

[0048] Figure 6 It is the relative error diagram of adjacent vehicle systems in Embodiment 1 of the present invention. Specific embodiments

[0049] The present invention will be further clarified below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the following specific embodiments are only used to illustrate the present invention and not to limit the scope of the present invention.

[0050] Embodiment 1

[0051] An observer-based fault detection method for multi-vehicle systems, as Figure 1 shown, includes the following steps:

[0052] Step S1, Modeling of the vehicle system: Establish a leader-follower multi-vehicle model, model the dynamics of different vehicle systems, generate the parameter matrix of the system, and at the same time establish the communication network of the vehicle, characterize the topology matrix, the topology matrix is an undirected connected graph, and there is an edge connection between the leader and all followers. The topology matrix is also used for the design of the exponential observer in the subsequent step S2 and the computer simulation in step S4.

[0053] Mathematically model the leader of the vehicle and the vehicle to be detected (follower), specifically: Establish the dynamic equation of the vehicle system. As shown in formula (1):

[0054]

[0055] where v(t) represents the state of the leader of the vehicle system, y 0 (t) represents the output value of the leader, S is the leader system matrix, and R is the leader output matrix. For the follower, there is the following system dynamic equation

[0056]

[0057] where x k(t) represents the state of the k-th vehicle follower, y k (t) represents the output value of the follower, A k , B k , C k , E k are the parameter matrices of the follower system.

[0058] Next, group every three vehicles as a group, number the vehicles in each group for subsequent group detection. Select one of the groups and establish a network communication topology from the leader to the followers.

[0059] Step S2: Exponential observer design: Design an observer for the leader from the parameter matrices obtained in Step S1.

[0060] For the parameter matrices obtained by the leader in Step S1, design an exponential distributed observer for the leader system, which is specifically expressed as the following formula:

[0061]

[0062] where χ k (t) represents the estimated value of the state of the k-th vehicle for the leader at time t, v(t - h) represents the state of the leader at time t - h, y k (t), y j (t) represent the output values of the k-th and j-th vehicles respectively, h represents the communication time delay from the leader to the followers, γ is a parameter to be designed satisfying γ > 0, L is the observation gain matrix, S is the system matrix of the leader. If there is a signal connection between vehicle k and vehicle j, then h kj > 0, otherwise h kj = 0. Figure 2 The fault detection structure diagram of follower k based on the observer is given, Figure 3 The structure diagrams of three follower vehicles based on the distributed observer are given. It can be seen from the figure that the three follower vehicles realize supervision and alarm and implement distributed observation based on the relative measurement values through the detection mechanism.

[0063] Step S3: Computer simulation: Based on the leader system dynamics model in Step S1 and the observer designed in Step S2, give the estimation error system and the control protocol of the followers. For the system parameter model obtained in Step S1, through the Lyapunov stability theory and the small gain theorem, give the criterion for the output consensus of the multi-vehicle system. Based on the vehicle dynamics model in Step S1, set the initial values and the consensus criterion, and conduct numerical simulation to obtain the relative measurement of the follower vehicle system, and use the relative measurement for the design of the alarm mechanism in the subsequent Step S4. Specifically: Based on formulas (1) and (2), obtain the following observation error

[0064]

[0065] and the observation error system

[0066]

[0067] where e γ-1 ≥γ, d = -γe -γ e -d , d ∈ [-1, 0).

[0068] Based on the follower system in step S1, consider the following control protocol

[0069]

[0070] where u k (t) represents the control input of the k-th vehicle, is the forward gain of the k-th vehicle, K xk is the feedback gain of the k-th vehicle.

[0071] To achieve consensus, the following assumptions need to be satisfied:

[0072] Assumption 1: The network topology between the leader and the followers is undirected and connected, and there is an edge connecting the leader and all the followers.

[0073] Assumption 2: There exist matrices Π k and U k satisfying

[0074]

[0075] where A k B k , C k , E k , k = 1, …, N are the parameter matrices of the follower system, S is the leader system matrix, and R is the leader output matrix.

[0076] Based on the Lyapunov stability theory and the small gain theorem, the following criterion for the output consensus of the multi-vehicle system is obtained: Considering formula (1) of the leader system, formula (2) of the follower system, the observer formula (3), the control protocol (5), and assuming that both Assumption 1 and Assumption 2 are satisfied, if there exist a time delay h and a positive definite matrix P satisfying the following formula

[0077]

[0078] then there exists a control gain matrix such that formula (1) and formula (2) of the vehicle system achieve output consensus, where, is the identity matrix, Let \(d\) be a real number such that \(d\in[-1,0)\), \(S\) be the leader system matrix, the matrix \(S\) is normal, \(\mu(S)\) be the largest positive real part of the leader matrix \(S\), and \(I\) n be the \(n -\)dimensional identity matrix, and \(e\) denote the Euler number.

[0079] According to the established criterion above, computer simulation of the vehicle system is carried out to obtain the measured values of the adjacent vehicle (follower) system, which serves as the input item for the fault detection algorithm design in the subsequent step S4.

[0080] The following set of vehicle systems is taken as an example in this implementation:

[0081] Set the parameter matrix of the leader system as

[0082]

[0083] The parameter matrix of the follower system is

[0084]

[0085] where \(a\) k , \(b\) k , \(d\) k are real numbers, \(k = 1,2,3\). The network communication topology is as Figure 4 shown, and the Laplacian matrix

[0086]

[0087] From the output regulation equation (6), it can be obtained that:

[0088]

[0089] where \(k = 1,2,3\).

[0090] Set the parameters as \(a\) k , \(b\) k , \(d\) k =\(\{3,1,1\}\), \(\{1,2,1\}\), \(\{2, - 1,1\}\), and further set the initial values of each vehicle system as \(v(0)=[0.2\ 0.5]'\), \(x\) 1 (0)=[0.2,0.3]'\), \(x\) 2 (0)=[0.1,0.6]'\) and \(x\) 3 (0)=[0.5,0.2]'\), the time delay \(h = 2\), \(L = 10\) -6 [1\ 0], \(\gamma=0.0793\). Select \(K\) xk such that \(A\) k +\(B\) k \(K\) xk is a Hurwitz matrix.

[0091] Then there is \(K\)x1 = [-1.33 -4], K x2 = [-1.47 -1.75], K x3 = [3.045 5]. Based on the formula The following control gain matrix is further obtained . Then, computer simulation is carried out on the vehicle system (leader-follower system) to obtain the maximum value of the measurement difference norm of adjacent vehicle systems, which is used as an input item for the fault judgment mechanism in step S4. Figure 5 It shows that the leader-follower system in this embodiment realizes an output consistent trajectory curve.

[0092] Step S4, alarm detection design: Based on the relative measurement values in the computer simulation in step S3, a judgment mechanism is introduced, and an alarm algorithm is designed to enable real-time fault detection of the vehicle system during operation to determine the fault condition of the follower vehicle.

[0093] From the relative measurement values obtained in step S3, the alarm mechanism is set as: ||y k - y j | ≤ σ, k, j are adjacent vehicles, σ is the alarm threshold, and during the actual operation of the vehicle, if the relative measurement value exceeds the threshold σ (judged by the discrimination mechanism), an alarm is given.

[0094] Select a set of detection mechanism algorithms for the vehicle system, and the other groups of vehicles are analogized. According to step S1, the vehicles are numbered. Three vehicles are numbered from left to right, and the vehicles are numbered as vehicle 1, vehicle 2, and vehicle 3 respectively. Denote the alarm thresholds of vehicles 1, 2, and 3 as σ 1 , σ 21 , σ 23 , σ 3 . Among them, vehicle 2 has two threshold values σ 21 , σ 23 , and as long as one of them is triggered, an alarm is given. As Figure 6 shown, we can obtain that the alarm threshold of vehicle 1 is σ 1 = 1.5, the alarm threshold of vehicle 2 is σ 21 = 1.5 and σ 23 = 1.9, and the alarm threshold of vehicle 3 is σ 33 = 1.9. Further, the following alarm mechanism algorithm is obtained:

[0095] If vehicle 1 and vehicle 2 give an alarm (the adjacent measurement values exceed the given threshold) and vehicle 3 does not give an alarm, it means that vehicle system 1 has a fault.

[0096] If vehicle 3 and vehicle 2 give an alarm and vehicle 1 does not give an alarm, it means that vehicle 3 has a fault.

[0097] If Vehicle 1, Vehicle 2 alarms and Vehicle 3 alarms, it indicates that there is a fault in Vehicle 2 or faults occur in all of Vehicle 1, Vehicle 2 and Vehicle 3.

[0098] In summary, in a multi-vehicle system, due to the limitation of the space structure, when designing the alarm mechanism, it is often necessary to save space. To make better use of information and space resources, a detector based on a distributed observer is introduced to achieve real-time supervision and detection of vehicles. While achieving vehicle system consistency, the present invention performs real-time supervision of vehicles based on the observer-based fault mechanism, ensuring the safety of the vehicle system, reducing the conservativeness of the upper bound of communication time delay, and providing a new idea for the research on the safety detection of multi-vehicle systems.

[0099] It should be noted that the above content only illustrates the technical idea of the present invention and cannot be used to limit the protection scope of the present invention. For those of ordinary skill in the art of this technology, without departing from the principle of the present invention, several improvements and refinements can still be made, and these improvements and refinements all fall within the protection scope of the claims of the present invention.

Claims

1. A multi-vehicle system fault detection method based on an observer, characterized in that: The steps include: S1. Establish vehicle system model: mathematically model the navigator and follower of the vehicle, establish the dynamic equation of the vehicle system, generate the parameter matrix of the system, and establish the network communication topology from the navigator to the follower; S2. Exponential distributed observer design: Based on the parameter matrix obtained in step S1, an exponential distributed observer is designed for the navigator; S3, computer simulation: according to the dynamic equation of the vehicle system in step S1 and the observer in step S2, the observation error and the observation error system are calculated; based on Lyapunov stability theory and the small gain theorem, the observation gain and the control gain are set, the criterion for the consistent output of the multi-vehicle system is determined, and the simulation is performed to obtain the relative measurement values ​​of adjacent vehicles; the criterion is specifically: if there is a time lag h and the positive definite matrix P satisfies the following formula: Then there exists a control gain matrix K xk , K xk , so that the vehicle system dynamics equation and the follower dynamics equation achieve consistent output; Where I is the identity matrix, d is a real number satisfying d∈[-1,0), S is the navigator system matrix, the matrix S is normal, μ(S) is the maximum positive real part of the matrix S, I n is the n-dimensional identity matrix, e represents the Euler number; S4, alarm detection: set a threshold value for the relative measurement value of the adjacent vehicles obtained in step S3. When the relative measurement value of the vehicle is greater than the threshold value during actual vehicle operation, an alarm is triggered to determine that there is a fault in the vehicle system, thereby achieving fault detection.

2. The observer-based multi-vehicle system fault detection method according to claim 1, characterized in that: In step S1, the navigator dynamics equation of the vehicle system is specifically: y0(t)=Rv(t) in represents the navigator state of the vehicle system, y0(t) represents the output value of the navigator, S is the navigator system matrix, and R is the navigator output matrix; The dynamic equation of the follower system is specifically: y k (t)=C k x k (t),k=1,…,N in represents the state of the k-th vehicle follower, y k (t) represents the output value of the follower, A k , B k , C k E k is the parameter matrix of the follower system.

3. The multi-vehicle system fault detection method based on distributed observer as claimed in claim 2, characterized in that: The distributed observer in step S2 is specifically: where χ k (t) represents the state estimate of the leader by the kth follower at time t, v(th) represents the state of the leader at time th, and y k (t), y j (t) represents the output value of the kth and jth vehicles, respectively, h represents the communication delay from the leader to the follower, γ is a parameter, satisfying γ>0, L is the observation gain matrix, S is the leader system matrix, if there is a signal connection between vehicle k and vehicle j, then h kj >0, otherwise h kj =0.

4. The observer-based multi-vehicle system fault detection method according to claim 3, characterized in that: The computer simulation process of step S3 satisfies the following conditions: the network topology of the navigator and the followers is undirectedly connected, there is an edge connecting the navigator and all the followers, and there is a matrix Π k and U k satisfy P k S=A k P k +B k U k +E k 0=C k P k -R, k=1,…N Among them A k , B k , C k , E k , k = 1,…, N is the system matrix of the follower, S is the system matrix of the leader, and R is the output matrix of the leader.

5. The observer-based multi-vehicle system fault detection method according to claim 4, characterized in that: The step S3 follower system includes a control protocol: where u k (t) represents the control input of the kth vehicle, is the forward gain of the kth vehicle, K xk is the feedback gain of the kth vehicle.

6. The observer-based multi-vehicle system fault detection method according to claim 5, characterized in that: In step S1, the vehicles in the followers are classified into groups of three and numbered sequentially, namely vehicle 1, vehicle 2 and vehicle 3. The alarm thresholds of vehicles 1, 2 and 3 are σ1, σ 21 , σ 23 , σ3, when an alarm occurs in step S4, the specific fault judgment rules are as follows: If vehicle 1 and vehicle 2 give an alarm, but vehicle 3 does not, then vehicle 1 has a fault; If vehicles 2 and 3 give an alarm, but vehicle 1 does not, it means that vehicle 3 has a fault; If vehicle 1, vehicle 2 and vehicle 3 alarm, it means that vehicle 2 breaks down or vehicle 1, vehicle 2 and vehicle 3 all break down.

7. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and the computer program is executed by a processor to implement the observer-based multi-vehicle system fault detection method as claimed in any one of claims 1 to 6.

8. A computer device, characterized in that: include: A memory for storing instructions; A processor is used to execute the instructions so that the computer device performs the operation of the observer-based multi-vehicle system fault detection method as described in any one of claims 1 to 6.

Images