Multi-vehicle platoon control method and system against cyber attacks

Abstract

The application discloses a kind of anti-network attack's multi-vehicle safety platoon control method and system, the method first establishes multi-vehicle queue dynamics model;Second, design fuzzy state observer to estimate vehicle state and unknown nonlinear dynamics;Then, introduce funnel function to build time-varying performance boundary, and the double constraints of vehicle anti-collision and keep communication are converted into funnel constraints of spacing tracking error;Then, based on backstepping method and funnel performance error, design virtual controller and adaptive law, and obtain adaptive safety controller to obtain control input;Finally, through Lyapunov stability analysis, the tolerance threshold condition that system can still maintain final consistent bounded stable after suffering replay attack is quantitatively obtained.The application simplifies design through funnel control, avoids complex barrier lyapunov function, and for the first time clearly provides quantified anti-interference ability to replay attack, enhances the network security and practicality of multi-vehicle queue system.

Description

Technical Field

[0001] This invention relates to the field of multi-vehicle safe constrained platooning control, and in particular to a multi-vehicle safe platooning control method and system resistant to network attacks. Background Technology

[0002] In recent years, with the rapid iteration and upgrading of wireless communication and automation technologies, the number of motor vehicles in urban traffic has grown exponentially. This has led to increasingly prominent problems such as traffic congestion, frequent accidents, and excessive pollutant emissions. Therefore, intelligent transportation systems (ITS) have developed rapidly to effectively alleviate these challenges. As a core component of ITS, multi-vehicle platooning has attracted widespread attention from researchers worldwide. From a physical perspective, maintaining appropriate vehicle spacing is crucial for ensuring the safety of vehicle platooning (i.e., avoiding collisions) and maintaining reliable communication connections. If the spacing is too large, communication may be interrupted due to the limited detection range of onboard sensors, preventing effective information transmission; conversely, if the spacing is insufficient, collisions may occur between adjacent vehicles in emergency braking situations. Therefore, the problems of collision avoidance and stable communication connections have attracted the attention of many control theory and control engineering researchers, resulting in a series of research achievements. From a network perspective, information transmission between vehicles in a multi-vehicle platooning system relies on the communication network, making the platooning system vulnerable to cyberattacks, which can affect the system's stable operation.

[0003] In the field of vehicle control, multi-vehicle platooning systems can ensure a safe distance between adjacent vehicles, thereby reducing traffic accidents that may occur during emergency braking. Currently, although many control technologies are used in vehicle platooning control systems, existing technologies still have the following problems: First, in the control methods of vehicle queuing systems, although existing technologies can avoid vehicle collisions, most of them require the design of complex obstacle Lyapunov functions.

[0004] Second, most existing security control methods only consider the impact of DoS attacks or FDI attacks on multi-vehicle queues, without considering the system's tolerance to replay attacks. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention combines a safe formation control method with a funnel function to propose a safe formation controller structure and design method based on a funnel function, thereby achieving safe formation control for a networked multi-vehicle queuing system.

[0006] To achieve the above objectives, the technical solution of the present invention is as follows: a multi-vehicle secure platooning control method and system resistant to network attacks. The control method comprises a fuzzy state observer module, a funnel constraint module, a virtual controller module, an adaptive law module, a replay attack tolerance analysis module, and a multi-vehicle platooning system. The input of the fuzzy state observer module is connected to the output of the multi-vehicle platooning system; the output of the fuzzy state observer module is connected to the input of the funnel constraint module; the input of the virtual controller module is connected to the output of the funnel constraint module; the output of the virtual controller module is connected to the input of the adaptive law and the input of the replay attack tolerance analysis module; the output of the adaptive law module is connected to the input of the virtual controller module; and the output of the replay attack tolerance analysis module is connected to the input of the multi-vehicle platooning system.

[0007] To achieve the above objectives, the technical solution of the present invention is as follows: A. Establish a multi-vehicle queuing system model The multi-vehicle queuing system model is described as follows: (1)

[0008] in, For the defined intermediate variables, , , The first The vehicle's position, speed, and acceleration To control the input, define intermediate variables. ,in For the first The quality of the vehicle For the first The vehicle's hysteresis coefficient, For an unknown continuous nonlinear function, This is the system output.

[0009] The leader model of the multi-vehicle queuing system is described as follows: (2) in, , , For the leader's position, speed, and acceleration.

[0010] B. Fuzzy State Observer Module The main function of the fuzzy state observer module is to handle the problem of unpredictable system state in actual systems and obtain an estimate of the system state through the fuzzy state observer.

[0011] Based on the approximation capability of fuzzy logic systems, to approximate unknown dynamic parameters, the fuzzy state observer is designed as follows: (3) Among them, variables , , They represent , , The estimated value, yes The estimated value of the variable for The estimated value, and satisfying , For positive integers, , , For observation gain, These are known fuzzy basis functions.

[0012] According to fuzzy logic systems, unknown nonlinear dynamics satisfy the following approximation relationship: (4) in, It is an ideal weight vector. It is a positive constant and satisfies , Represents positive integers, and yes Continuous, and satisfying the following inequalities: (5) in, Is with Related known information constant.

[0013] C. Funnel Constraint Module In actual driving, to ensure the safety of each vehicle in a multi-vehicle platooning system, safe distances and communication constraints must be considered. Therefore, the main function of the funnel constraint module is to define the following safety, communication connectivity, and performance constraints.

[0014] (1) Vehicle safety distance and communication connection constraints The distance between adjacent vehicles is defined as: (6) in, For the first The length of the vehicle.

[0015] To ensure safe vehicle distances and communication connectivity constraints, the distance between adjacent vehicles is limited as follows: (7) in, The minimum safe distance to avoid a collision. Indicates the maximum communication connection distance.

[0016] The expected distance between adjacent vehicles is: (8) in, It is the expected distance between adjacent vehicles when the vehicle is stopped. It is the expected distance. It is a constant time interval.

[0017] According to formulas (7) and (8), the spacing error can be obtained as follows: (9) in, satisfy , and yes The known minimum and maximum limits.

[0018] (2) Performance constraints Spacing error is ensured by using funnel control. The performance constraints must be met. Consider the following funnel boundary. and performance funnel set: (10) According to formula (10), we can obtain: (11)

[0019] Through design Constraint (11) includes the previous constraint (9). Essentially, satisfying constraint (11) is sufficient to guarantee that constraints (7) and (9) hold.

[0020] D. Virtual Controller Module The main function of the virtual controller module is to derive the virtual controller of the system from the nonlinear model of the multi-vehicle queuing system.

[0021] First, the following error transformation is performed:

[0022] (12) in, For virtual controllers, A constant time interval coefficient, This represents the expected distance when adjacent vehicles stop.

[0023] Based on the funnel constraints, the funnel error design can be obtained as follows: (13) in, , It is a funnel boundary function. , This is the performance function of the funnel.

[0024] The virtual controller then takes the form of: (14)

[0025] in, It is a constant. , for The time derivative.

[0026] E. Adaptive Law Module The function of the adaptive law module is to calculate the dynamic changes of the adaptive parameters, and then send the calculated dynamic changes of the adaptive parameters to the virtual controller module.

[0027] Introduce an adaptive law for the adaptive parameters in a multi-vehicle queuing system: (15) in, For positive integers, .

[0028] F. Replay Attack Tolerance Analysis Module The main function of the replay attack tolerance analysis module is to detect the system's tolerance to replay attacks, that is, if the system error changes within a limited threshold after the system suffers a replay attack, the system can remain stable.

[0029] The controller design for the multi-vehicle queuing system is as follows: (16) in, It is a constant.

[0030] Each ( All of these can be represented as follows: , and Linear combination: (17) in, , , , All are constants. To simplify the analysis, we define... and We can obtain: (18) exist In the case where the error caused by the attack satisfies the following condition, the nonlinear vehicle queuing system provided by (1) and controlled by (16) is eventually uniformly bounded: (19) in, (20) (twenty one) (twenty two) in, .

[0031] Compared with the prior art, the present invention has the following beneficial effects: First, by introducing a time-varying funnel boundary function to directly constrain the spacing tracking error, the complex dual constraints of collision avoidance and communication maintenance are unified into simple performance constraints. This avoids the tedious process of constructing complex obstacle Lyapunov functions in traditional methods, and significantly simplifies controller design and analysis.

[0032] Second, it not only considers the existence of replay attacks, but also, through rigorous stability analysis, derives for the first time a clear tolerance threshold condition to ensure the stability of a multi-vehicle queuing system under replay attacks, providing quantitative indicators for system security assessment.

[0033] Third, the proposed modular controller structure (observer-funnel constraint-virtual control-adaptive law-attack analysis) has a clear hierarchy, with each module having a well-defined function, making it easy to implement and integrate on the vehicle-mounted computing unit. Attached Figure Description

[0034] This invention has a total of appendices Figure 6 Zhang, of which: Figure 1 This is a schematic diagram of a vehicle safety formation controller based on a funnel function.

[0035] Figure 2 It is a displacement trajectory diagram of the three following vehicles and the lead vehicle.

[0036] Figure 3 It shows the speed trajectory of the three following vehicles and the lead vehicle.

[0037] Figure 4 It shows the acceleration trajectory of the three following vehicles and the lead vehicle.

[0038] Figure 5 It is a trajectory diagram showing the spacing error between the three following vehicles.

[0039] Figure 6 It is a control signal trajectory diagram of 3 following vehicles. Detailed Implementation

[0040] The invention will now be described in further detail with reference to the accompanying drawings. The structure of a multi-vehicle queue safety formation controller based on a funnel function designed in this invention is as follows: Figure 1 As shown. During vehicle movement, the vehicle displacement... The information is input into the fuzzy state observer module to obtain the state estimate. , , The data is input to the funnel constraint module for constraint processing. After being constrained by the funnel constraint module, the data is then input to the virtual controller module to obtain the virtual controller. The virtual controller obtained from the virtual controller module Adaptive parameters in the adaptive law module The input is fed into the replay attack tolerance analysis module, and the final generated control signal is then transmitted back to the multi-vehicle queuing system. The design goal of this invention is to effectively reduce the amount of computation by introducing a funnel function into the control design process of the multi-vehicle queuing system, thereby further avoiding vehicle collisions and enabling communication and performance constraints on the vehicles.

[0041] To enable those skilled in the art to implement the present invention, a specific simulation embodiment is described below.

[0042] 1. Simulation Scenarios and Parameter Settings This embodiment considers a straight convoy consisting of one lead vehicle and three follower vehicles (N=3). The vehicle parameters are as follows: Navigator: Initial position:

[0043] Initial velocity:

[0044] Acceleration: Train Commander:

[0045] Following vehicle (i=1,2,3): Mass:

[0046] Hysteresis coefficient:

[0047] Initial position:

[0048] Initial velocity:

[0049] Initial acceleration:

[0050] Vehicle length:

[0051] Safety constraints:

[0052] 2. Specific implementation parameters of the controller module: fuzzy state observer module (corresponding to formula 3): the observer gain is set to... k 1,i , k 2,i , k 3,i Fuzzy basis functions Choose the Gaussian function.

[0053] Fuzzy basis functions: ,in, ; Observation gain:

[0054]

[0055]

[0056] Funnel constraint module (corresponding to formulas 10 and 11): The funnel boundary is designed in exponential form to ensure that the error converges quickly and is always constrained, specifically as follows:

[0057] Virtual controller module (corresponding to formula 14): Design constants:

[0058] Adaptive law module and final control law (corresponding formulas 15 and 16): Final controller gain

[0059] Adaptive law parameters:

[0060] 3. Simulation Results and Analysis A closed-loop system model was built and simulated in the MATLAB / Simulink environment. The results are as follows: Figures 2 to 6 As shown.

[0061] Normal operating performance: Figure 2 , Figure 3 , Figure 4 The displacement, velocity, and acceleration trajectories of the three following vehicles and the lead vehicle are shown respectively. It can be seen that the following vehicles can accurately and smoothly track the movement of the lead vehicle. Figure 5 The distance error of the three following vehicles was shown. Trajectory. Crucially, all error curves are constrained throughout by... ( tWithin the defined funnel boundary, the effectiveness of the invention in avoiding collisions (error does not fall below the lower limit) and maintaining communication (error does not fall above the upper limit) is intuitively demonstrated. Figure 6 Display control input signals ( t It is continuous and physically realizable.

[0062] Replay Attack Verification: To verify the system's tolerance to replay attacks, multiple replay attacks were performed on the system. For example, during the period from t=15s to 18s, all following vehicle controllers were subjected to replay attacks. Injection cycle A replay attack with a duration of 3 seconds and within a threshold was observed during simulation. The simulation revealed that under this attack, the convoy spacing error ( Figure 5 During the attack, bounded fluctuations occurred, but the funnel boundary was never violated, and the convoy maintained stable formation without collisions. This empirically demonstrates the correctness of the threshold condition proposed in this invention, as well as the system's resistance to attacks within the quantified security range.

[0063] In summary, this specific embodiment, through detailed parameter settings and simulation verification, fully demonstrates that the control method and system provided by this invention not only achieve integrated protection of security and communication performance through funnel constraints, simplifying the design, but more importantly, it can quantitatively assess and resist replay attacks, significantly improving the practical security and robustness of the connected multi-vehicle queuing system.

Claims

1. A method for secure multi-vehicle platooning control resistant to network attacks, characterized in that, Includes the following steps: S1. Establish a dynamic model for a multi-vehicle platoon system: The platoon consists of one lead vehicle and N following vehicles; for the i-th following vehicle, its dynamic model is as follows: ； in, , , Let be the position, velocity, and acceleration of the i-th vehicle, respectively. To control the input, For state vectors, Given parameters, For vehicle quality, The hysteresis coefficient, For an unknown continuous nonlinear function, For system output; S2. State and parameter estimation based on fuzzy logic system: Design a fuzzy state observer and use the fuzzy logic system to approximate the unknown nonlinear function. And estimate the speed of the following vehicle online. acceleration and weight vector ; S3. Performance constraint design based on funnel function: Define the actual distance between the i-th vehicle and the vehicle in front. ,in The length of the vehicle in front; based on the minimum safe distance for collision avoidance. and maintain the maximum communication distance Construct time-varying funnel boundary functions This causes the spacing tracking error to be... Constrained within the preset funnel boundary, i.e., satisfying ,in The desired distance; S4. Design of virtual control law based on funnel performance error: using the state estimate obtained in step S2 , Define transformation error , and the corresponding funnel performance error , Design virtual control laws The virtual control law is the transformation error. Funnel performance error Funnel boundary function Functions of its derivatives; S5. Design an adaptive law to dynamically update the estimated weight vector. Combined with the aforementioned virtual control law Its derivative The transformation error Funnel performance error Funnel boundary function and its derivative and weight vector estimate To obtain adaptive safety control input (t); S6. Model the replay attack as a repeated injection of historical data from the system state or control channel; derive the threshold condition for the replay attack strength that maintains the eventual consistent bounded stability of the closed-loop system through Lyapunov stability analysis; the threshold condition is expressed as the change in system parameters, controller parameters, and the error caused by the attack. Inequalities.

2. The method according to claim 1, characterized in that, In step S2, the design of the fuzzy state observer is based on the following approximation relation: ,in For the ideal weight vector, Given a known fuzzy basis function vector, To meet The approximation error, Represents positive constants; the specific form of the observer is: ； in, , , They are respectively , , The estimated value, yes The estimated value of the variable for The estimated value, and satisfying , For positive integers, = , , , For observation gain, =[ , ] T .

3. The method according to claim 2, characterized in that, In step S4, the transformation error is defined as: ； ； in, A constant time interval coefficient, The desired distance when the vehicle stops; the funnel performance error is defined as: ； in, , For the corresponding funnel boundary function, , This is the performance function of the funnel.

4. The method according to claim 3, characterized in that, In step S5, the adaptive law is designed as follows: ； in, It is an adjustable parameter. The adaptive safety control input (t) is designed as follows: ； in, For controller gain, .

5. The method according to claim 4, characterized in that, In step S6, the replay attack strength threshold condition is specifically as follows: ； in, Indicates the attack time as The cumulative state error change caused by the replay attack. ； in, It is a fuzzy basis function vector The upper bound of the Lipschitz constant, It is the ideal weight vector upper bound of the norm, , These are the upper bounds for the coefficients of the nonlinear terms in the system model and the coefficients of the coupling terms in the controller, respectively. and Representing the current time and The funnel performance error of the i-th vehicle at time i.

6. A network-attack-resistant multi-vehicle safe platooning control system for implementing the method of any one of claims 1 to 5, characterized in that, This includes a physical multi-vehicle queuing system and a controller that interacts with it; the controller logically integrates the following functional modules: Fuzzy State Observer Module: Used to perform the state and parameter estimation function as described in claim 2; Funnel constraint and performance error generation module: used to perform the distance calculation, error calculation and funnel boundary generation functions in step S3 of claim 1, and output the funnel performance error as described in claim 3. , ; Virtual controller module: used for performance error of the funnel And the funnel boundary function, calculate the virtual control law as described in claims 1 and 3. ; Adaptive law module: used to adjust the funnel performance error accordingly. And the fuzzy basis function vector, execute the adaptive law as described in claim 4, and update the approximation weight vector. ; Adaptive security control law synthesis and replay attack analysis module: used to integrate the virtual control law and its derivative, the output of the adaptive law module and the performance error of the funnel Synthesize the adaptive safety control input as described in claim 4 (t); Meanwhile, the module has a built-in algorithm to evaluate the replay attack tolerance of the system in its current state, and the evaluation is based on the threshold condition as described in claim 5.

7. The system according to claim 6, characterized in that, The functions of the fuzzy state observer module, funnel constraint and performance error generation module, virtual controller module, adaptive law module, and adaptive safety control law synthesis and replay attack analysis module are implemented in the onboard computing unit of each following vehicle through software algorithms; each onboard unit exchanges the state of the navigator vehicle through a wireless communication network. , , It also obtains the necessary status information of adjacent following vehicles to collaboratively complete formation control.

Images