Intelligent vehicle cluster safe formation control method based on observer

Abstract

The invention discloses an intelligent vehicle cluster safe formation control method based on an observer, an intelligent vehicle cluster comprises a leader vehicle and follower vehicles, and a third-order longitudinal dynamics mathematical model is constructed for each follower vehicle and each leader vehicle. A sampling-based distributed observer is constructed based on a communication relationship among vehicles and a third-order longitudinal dynamics mathematical model, and an extended state observer is constructed based on state information of a follower vehicle and introduction of unmodeled disturbance; an intelligent vehicle cluster safe formation control strategy is constructed based on a distributed observer and an extended state observer, and a DoS attack model is constructed. According to the technical scheme, precise estimation of the state information of the leader vehicle by the follower vehicles and real-time precise estimation of each state component and unmodeled disturbance of each follower vehicle are achieved, it is guaranteed that each vehicle in the vehicle formation runs at a safe interval and at a constant vehicle speed, and the robustness of a vehicle system in a complex environment is improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the cross field of intelligent transportation system and vehicle control engineering, and particularly relates to an intelligent vehicle cluster safety platoon control method based on an observer. BACKGROUND

[0002] With the development of the automobile industry, the number of vehicles on the road has increased dramatically, the road traffic efficiency has decreased, traffic accidents have occurred frequently, and the traffic situation has deteriorated. In view of this, the intelligent traffic system (ITS) has ushered in rapid development, and the intelligent vehicle platoon control technology has gradually become a key means to improve road traffic efficiency and enhance road traffic safety. Vehicle platoon technology realizes fixed or variable safety distance platoon driving through multi-vehicle cooperation, which can effectively save road space, relieve traffic pressure, reduce vehicle energy consumption, and provide key technical support for automatic driving technology.

[0003] Vehicle platoon control faces many challenges in complex dynamic traffic scenarios. First, vehicle platoon control is extremely dependent on vehicle networking technology (V2V / V2X) to realize the exchange of state information between vehicles. However, in actual communication scenarios, due to limited communication resources, obstruction blocking, and cost constraints, the communication topology of vehicle platoon is difficult to maintain a fully connected state. Therefore, in most cases, the communication topology is in a sparse state, which makes some vehicles unable to accurately obtain leader state information due to information loss, thereby causing platoon instability.

[0004] Secondly, the uncertainty and variability of actual traffic scenarios cause vehicles to be subjected to extremely complex dynamic disturbances during driving, such as wind resistance changes (proportional to the square of the vehicle speed), slope resistance, mechanical disturbances, and load disturbances, etc. These factors bring uncertainty to vehicle modeling. Existing traditional model predictive control (MPC), PID control, etc. methods all require accurate disturbance models, however, the disturbance in actual traffic scenarios has strong time-varying and unmeasurable characteristics, which makes it difficult for vehicle control to accurately compensate for the disturbance, causing vehicle acceleration tracking error accumulation, and thus threatening platoon safety.

[0005] In addition, during vehicle platoon control, the communication between vehicles is usually in an open network environment, and the open communication architecture of vehicle networking is vulnerable to malicious network attacks. Existing platoon control strategies lack flexibility design against network attacks, and once the communication is interrupted, the queue may be disintegrated. Therefore, it is necessary to verify the designed control strategy under network attack environment to ensure that it can still be stably and effectively controlled when subjected to network attack. SUMMARY

[0006] To overcome the shortcomings of existing technologies, this invention provides an observer-based method for safe formation control of intelligent vehicle clusters.

[0007] This invention is achieved through the following technical solution: A method for safe platooning control of intelligent vehicle clusters based on observers includes the following steps: S1: The intelligent vehicle cluster consists of one leader vehicle and... One follower vehicle, >0; For a single following vehicle, construct a third-order longitudinal dynamic mathematical model of the following vehicle, which includes the three-dimensional state vector of the following vehicle. , , , , They represent the first The position, speed, and acceleration of each follower vehicle. For time; construct a third-order longitudinal dynamic mathematical model of the leader, which includes the three-dimensional state vector of the leader vehicle. , , , These represent the position, speed, and acceleration of the leader's vehicle, respectively. S2: Constructing inter-vehicle communication relationships for intelligent vehicle clusters; S3: Based on inter-vehicle communication relationships, a third-order longitudinal dynamics mathematical model of the follower, a third-order longitudinal dynamics mathematical model of the leader, and the first observation gain matrix. Second observation gain matrix Regarding the first Each follower vehicle constructs a sampling-based distributed observer to obtain the first... A follower vehicle The estimated value ;based on and Construct a distributed observer to estimate the error; S4: Targeting the A follower vehicle, constructing a four-dimensional state vector. , Indicates unmodeled disturbances, based on Construct an extended state observer, and obtain the results through the extended state observer. The estimated value , , , , They are respectively , , , The estimated value; based on and Construct an extended state observer to estimate the error; S5: Definition of the first The safe distance between the follower vehicle and the leader vehicle; for the first A follower vehicle, based on a distributed observer, an extended state observer, a safety interval, and a control gain matrix. Construct a safe platooning control strategy for intelligent vehicle clusters; based on the safe platooning control strategy and safe intervals, construct the first... The tracking error of the safe distance between the follower vehicle and the leader vehicle; S6: Construct a DoS attack model; construct a segmented distributed observer estimation error under a DoS attack based on the DoS attack model and the distributed observer estimation error; construct a safe distance tracking error under a DoS attack based on the segmented distributed observer estimation error, the extended state observer estimation error, and the safe distance tracking error. S7: Construct a quadratic Lyapunov function based on distributed observer estimation error, extended state observer estimation error, safe distance tracking error, and a DoS attack model. Solve the control gain matrix using the Lyapunov function. First observation gain matrix Second observation gain matrix ; S8: Control of intelligent vehicle clusters is achieved by combining distributed observers, extended state observers, DoS attack models, and intelligent vehicle cluster security formation control strategies.

[0008] Further, in step S1, the third-order longitudinal dynamic mathematical model of the follower is: (1) in, , , Indicates the time delay of the dynamic system; Indicates the first The control input for a follower vehicle Indicates the first Unmodeled disturbances of individual follower vehicles; The third-order longitudinal dynamic mathematical model of the leader is as follows: (2) in, This represents the output value of the third-order longitudinal dynamics mathematical model of the leader. .

[0009] Furthermore, in step S2, vehicle-to-vehicle communication is achieved through V2V communication. The communication relationships between vehicles include those between follower vehicles and those between the leader vehicle and its followers. The communication relationships between follower vehicles form an undirected topology graph. ,in, Indicates the number of the following vehicle; It is an adjacency matrix. ,and If the first The following vehicle and the first If the following vehicles can communicate with each other, then ;otherwise, The communication relationship between the leader vehicle and the follower vehicles is as follows: If the first If a following vehicle can receive information from the leader, then ;otherwise, .

[0010] Further, in step S3, the distributed observer is: (3) in, Indicates the first A distributed observer following the vehicle Output value at time 10:00 and They represent the first The following vehicle and the first A distributed observer following the vehicle Output value at time 10:00 Indicates the number of samples. Indicates the sampling period; The estimation error of the distributed observer is: (4) Further, in step S4, the extended state observer is: (5) Combination Equation (4) can be restated as: (6) in, , , , Represents a zero matrix with 1 row and 3 columns; , Represents the observer gain factor. >0, and in the selection At this time, it is necessary to ensure that the real part of the configuration matrix eigenvalue is negative; The expansion state observer estimation error is: . (7) Further, in step S5, the intelligent vehicle cluster safety formation control strategy is: (8) Wherein, , , Indicates a 3-dimensional unit matrix, Indicates a zero matrix of 3 rows and 1 column; , Indicates the safety interval; The safety distance tracking error is: . (9) Further, in step S6, the DoS attack model includes: The First non-attack interval: ; The First attack interval: ; Wherein, Indicates the time when the First attack ends, Indicates the time when the First attack starts, Indicates the duration of the First attack; Because the attacker's energy is limited, the total attack duration And the attack frequency Of the DoS attack in the interval from any Time to Time is limited by the following conditions: , ; Wherein, , , , Is a normal number.

[0011] Further, in step S6, the segmented distributed observer estimation error is: (10) Wherein, , , wherein, Is the in-degree matrix, , ; Since the augmented state observer is not affected by the cyber attack, the augmented state observer estimation error of all follower vehicles can be expressed as: (11) where , , ; The safety distance tracking error under the DoS attack is: (12) where .

[0012] Further, in step S7, the quadratic Lyapunov function is: ; ; ; ; ; ; where , , , , , , , , , is a positive definite matrix, is an integral variable; The derivative of can be obtained. The distributed observer estimation error, the augmented state observer estimation error and the safety distance tracking error satisfy the global ultimate uniform bounded stability condition under the disturbance of unmodeled disturbance and DoS attack; The global ultimate uniform bounded stability condition is: where , , , , , , are all normal numbers, and the gain matrix , and can be obtained.

[0013] The beneficial effects that can be achieved by the present application are: 1. A distributed observer based on the inter-vehicle communication relationship and the third-order longitudinal dynamics mathematical model of each vehicle is constructed to realize accurate estimation of the state information of the leader vehicle by the follower vehicles.

[0014] 2. Based on the state information of each follower vehicle, an unmodeled disturbance is introduced to construct an extended state observer to realize real-time accurate estimation of each state component and unmodeled disturbance of each follower vehicle.

[0015] 3. Based on the safety interval between each follower vehicle and the leader vehicle, the distributed observer and the extended state observer, an intelligent vehicle cluster safety formation control strategy is constructed, which not only reduces the calculation burden related to data processing, but also compensates for unknown external disturbances to some extent, ensuring that each vehicle in the vehicle formation travels at a safe interval and a constant speed, effectively saving road space, relieving traffic pressure, and improving the robustness of the vehicle system in complex environments. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 It is a front-following (PF) topology diagram in the embodiment of the present application.

[0017] Figure 2 It is a double-front-following (TPF) topology diagram in the embodiment of the present application.

[0018] Figure 3 It is a lead-following-following topology (LTF) topology diagram in the embodiment of the present application.

[0019] Figure 4 It is an observation control diagram under DoS attack and external disturbance in the embodiment of the present application.

[0020] Figure 5 It is a position error component curve of the estimation error of the distributed observer in the embodiment of the present application.

[0021] Figure 6 It is a speed error component curve of the estimation error of the distributed observer in the embodiment of the present application.

[0022] Figure 7 It is an acceleration error component curve of the estimation error of the distributed observer in the embodiment of the present application.

[0023] Figure 8 It is an unmodeled disturbance and unmodeled disturbance estimation value curve of the first follower vehicle in the embodiment of the present application.

[0024] Figure 9 It is an unmodeled disturbance and unmodeled disturbance estimation value curve of the second follower vehicle in the embodiment of the present application.

[0025] Figure 10 Unmodeled disturbance for the 3rd follower vehicle in the embodiment of the present application, unmodeled disturbance estimation curve.

[0026] Figure 11 Position component curve of the safety distance tracking error for each follower vehicle in the embodiment of the present application.

[0027] Figure 12 Velocity component curve of the safety distance tracking error for each follower vehicle in the embodiment of the present application.

[0028] Figure 13 Acceleration component curve of the safety distance tracking error for each follower vehicle in the embodiment of the present application. DETAILED DESCRIPTION

[0029] An observer-based intelligent vehicle cluster safety formation control method, comprising the following steps: S1: the intelligent vehicle cluster comprises 1 leader vehicle and N follower vehicles, > 0; for a single follower vehicle, a follower third-order longitudinal dynamics mathematical model is constructed, which contains a three-dimensional state vector of the follower vehicle , , , , respectively represent the position, velocity and acceleration of the Nth follower vehicle, is time; a leader third-order longitudinal dynamics mathematical model is constructed, which contains a three-dimensional state vector of the leader vehicle , , , respectively represent the position, velocity and acceleration of the leader vehicle.

[0030] According to the force analysis of the car on the road and the car theory, the influence of the lateral and vertical directions can be ignored when the vehicle is in three-order longitudinal motion; when the vehicle is normally running, the transmission shaft deformation problem caused by acceleration can be ignored. Therefore, considering the engine driving force, the rolling resistance, the air resistance, the acceleration resistance and the slope resistance of the vehicle, the vehicle running equation can be obtained as: wherein, is the engine driving force, , is the engine torque, is the transmission ratio of the transmission, is the transmission ratio of the speed reducer, represents the mechanical efficiency, represents the wheel rolling radius; is the rolling resistance, , represents the road rolling resistance coefficient, represents the tire load; is the air resistance, , is the air resistance coefficient, is the wind area, is the air density, is the vehicle speed relative to the wind; is the acceleration resistance, , represents the vehicle gravity, represents the road slope angle; is the slope resistance, , is the rotational mass conversion coefficient, represents the vehicle mass, is the vehicle speed, is the acceleration.

[0031] The third-order longitudinal dynamics mathematical model of the follower is: (1) wherein, , , represents the time delay of the power system; represents the control input of the th follower vehicle, represents the unmodeled disturbance of the th follower vehicle.

[0032] The third-order longitudinal dynamics mathematical model of the leader is: (2) wherein, represents the output value of the third-order longitudinal dynamics mathematical model of the leader, .

[0033] S2: Construct the inter-vehicle communication relationship of the intelligent vehicle cluster.

[0034] The inter-vehicle communication is realized through V2V communication, and the inter-vehicle communication relationship includes the communication relationship between the follower vehicles and the communication relationship between the leader vehicle and the follower vehicles. The communication relationship between the follower vehicles is a undirected topological graph wherein, an index of a following vehicle; is an adjacency matrix, , and ; if the first following vehicle can communicate with the first following vehicle, ; otherwise, ; the communication relationship between the leader vehicle and the follower vehicle is if the first following vehicle can receive information of the leader, ; otherwise, .

[0035] S3: based on the communication relationship between vehicles, the third-order longitudinal dynamics mathematical model of the follower, the third-order longitudinal dynamics mathematical model of the leader, the first observation gain matrix and the second observation gain matrix , a sampling-based distributed observer is constructed for the first follower vehicle, and an estimated value of the first follower vehicle to is obtained through the distributed observer; a distributed observer estimation error is constructed based on and .

[0036] In a vehicle platoon, due to the limitations of the actual traffic environment and communication technology, not every follower vehicle can directly receive information of the leader vehicle. This makes part of the follower vehicles unable to accurately obtain the state information of the leader vehicle due to information loss, resulting in platoon instability. To solve this problem, the following distributed observer is designed, so that each follower vehicle can reduce the data processing requirements of vehicle control by estimating the leader state information.

[0037] The sampling-based distributed observer is: (3) wherein, represents the output value of the distributed observer of the first follower vehicle at time , and and represent the output values of the distributed observers of the first follower vehicle and the first follower vehicle at time , respectively, represents the sampling number, and represents the sampling period.

[0038] The distributed observer estimation error is: (4) represents a global vector containing the distributed observer estimation errors of all follower vehicles to the leader vehicle.

[0039] S4: for the i-th follower vehicle, construct a four-dimensional state vector , , represents an unmodeled disturbance, construct an extended state observer based on , obtain the estimated value of , , , , , , , , , , , , construct an extended state observer estimation error based on

[0040] Due to the uncertainty and variability of environmental factors in actual traffic scenes, vehicles will be disturbed by complex dynamic disturbances (which can be regarded as an unmodeled disturbance) during driving, so it is necessary to design an extended state observer to estimate the state of the follower vehicle and the dynamic disturbance it receives.

[0041] The extended state observer is: (5) Combined with , formula (4) can be re-expressed as: (6) wherein , , , represents a zero matrix of 1 row and 3 columns.

[0042] The extended state observer estimation error is: (7) is a global vector containing the extended state observer estimation errors of all follower vehicles.

[0043] Through the design of the above observer, the first purpose of the present application, i.e. the state estimation of the intelligent vehicle, has been achieved.

[0044] The sampling-based distributed observer enables each follower vehicle to accurately estimate the state information of the leader vehicle in a sparse communication topology scenario; the extended state observer accurately tracks and estimates the unmodeled dynamics, providing more abundant traffic environment data for subsequent control strategy design, making the control strategy design more scientific and reasonable.

[0045] S5: defining a safety interval between the first follower vehicle and the leader vehicle; for the first follower vehicle, constructing an intelligent vehicle cluster safe formation control strategy based on the distributed observer, the extended state observer, the safety interval, and a control gain matrix ; constructing a safety distance tracking error of the first follower vehicle to the leader vehicle based on the intelligent vehicle cluster safe formation control strategy and the safety interval. The intelligent vehicle cluster safe formation control strategy is:

[0046] The intelligent vehicle cluster safe formation control strategy is: (8) wherein, , , represents a 3-dimensional unit matrix, represents a zero matrix of 3 rows and 1 column; , represents the safety interval.

[0047] The safety interval usually needs to be designed with a corresponding spacing strategy. Different spacing strategies are selected and designed according to actual scene requirements: a constant spacing strategy (CS), in which the vehicle interval is usually designed as , is a constant, this spacing strategy is simple to implement, but has poor flexibility, and is suitable for low-speed and stable driving scenarios. A constant time interval strategy (CTH), in which the vehicle interval is usually designed as , is a constant, this spacing strategy is more suitable for changes in vehicle speed, such as highway driving, but requires high-precision vehicle-mounted sensors. A variable time interval strategy (VTH), in which the vehicle interval is usually designed as , is a constant, this spacing strategy can more flexibly adjust the safety distance according to the actual driving speed of the leader vehicle: in a low-risk scenario (such as uniform speed driving), the vehicle distance is actively reduced to improve road traffic capacity; and when detecting a sudden deceleration of the preceding vehicle or a sudden obstacle, the safety interval is automatically increased to reduce the risk of rear-end collision, but the dynamic time interval algorithm involves multi-parameter optimization, such as speed difference and acceleration feedback, which requires higher computing power of the vehicle-mounted processor.

[0048] The safety distance tracking error is: ​​ (9) This represents a global vector containing the tracking error of all follower vehicles relative to the leader at the safe distance.

[0049] The second objective of this invention has now been achieved, namely, the design of a safe platooning control strategy for intelligent vehicle clusters. This control strategy not only reduces the computational burden related to data processing, but also provides some compensation for unknown external interference.

[0050] S6: Construct a DoS attack model; construct a segmented distributed observer estimation error under a DoS attack based on the DoS attack model and the distributed observer estimation error; construct a safe distance tracking error under a DoS attack based on the segmented distributed observer estimation error, the extended state observer estimation error, and the safe distance tracking error.

[0051] Building a DoS attack model mainly involves designing and describing the attack interval, defining the time interval: No. Secondary non-attack zone: ; No. Secondary attack zone: ; in, Indicates the first At the end of the attack Indicates the first The moment the attack begins. Indicates the first The duration of the attack; Because attackers have limited resources, DoS attacks can occur in any... Time's up The total duration of all attacks within the time interval and attack frequency Subject to the following conditions: , ; in, , , , It is a normal number.

[0052] The estimation error of the piecewise distributed observer is: (10) in, ,in, Let be the in-degree matrix. , ; The extended state observer is not affected by cyber attacks due to its own design, and the estimation error of the extended state observer of all follower vehicles can be expressed as: (11) where ; since the energy of external disturbances received by the follower vehicles in the vehicle platoon is limited, the unmodeled disturbance and its first-order derivative are bounded.

[0053] The safety distance tracking error under DoS attacks is: (12) S7: Based on the distributed observer estimation error, the extended state observer estimation error, the safety distance tracking error under DoS attacks, and the DoS attack model, a quadratic Lyapunov function is constructed, and the control gain matrix , the first observation gain matrix , and the second observation gain matrix are obtained by solving the Lyapunov function.

[0054] The quadratic Lyapunov function is: ; ; ; ; ; ; where , , , , , , , , , are positive definite matrices, is an integral variable; Taking the derivative of can obtain the stability condition of the vehicle platoon system under the disturbance of unmodeled disturbance and DoS attacks to meet the global final consistent bounded stability condition, and , and are obtained by solving the stability condition, and , and ensure that the Lyapunov function satisfies: ; wherein, , , , , , , are all normal numbers.

[0055] S8: The control of the intelligent vehicle cluster is realized by combining the distributed observer, the extended state observer, the DoS attack model and the intelligent vehicle cluster safety formation control strategy.

[0056] In the embodiment of the present application, the establishment of inter-vehicle communication is realized by V2V (Vehicle-to-Vehicle) technology. V2V communication uses vehicle terminal equipment, so that vehicles can obtain real-time driving data of surrounding vehicles, including speed information, position coordinates and safety state parameters. In addition, the communication system can also build a multifunctional interactive platform to support instant transmission and sharing of text, image and video data between vehicles. For example, Figures 1-3 , the communication mode of V2V can be divided into front-following (PF) topology, double-front-following (TPF) topology and leader-following topology (LTF) topology according to the connection between homogeneous vehicles. It is worth mentioning that the present application is applicable to any of the above topologies.

[0057] The vehicle-mounted sensor adopts MEMS IMU (Micro-Electro-Mechanical Systems Inertial Measurement Unit). It is a kind of micro sensor device integrating accelerometer and gyroscope, which is used to measure the linear acceleration and angular velocity of the object in three-dimensional space, so as to obtain the motion attitude, position and direction information. Based on the sensor unit, the follower vehicle can obtain its real-time state information (i.e. the position, speed and acceleration information of the vehicle). However, due to the influence of external disturbance, these information needs to be more accurate state information estimation value through the extended state observer. Finally, the accurate follower state estimation value, disturbance estimation value and leader state information estimation value output by the distributed observer are simultaneously input into the controller, and the controller executes the intelligent vehicle cluster safety formation control strategy based on error feedback in the present application, further ensuring the accurate control and stable driving of the vehicle formation.

[0058] Experiment: In the simulation of vehicle formation safety control, a formation composed of 4 intelligent vehicles (including 1 leader vehicle and 3 follower vehicles) is designed, and the initial state is , , , In order to better simulate the unknown disturbance in the outside world, an energy-limited disturbance model is designed: The vehicle safety interval is designed as a constant according to the fixed spacing strategy, that is The observer gain and the control gain are obtained by further solving the sufficient condition obtained by Lyapunov stability analysis, which are , The simulation running time is 40s, and the simulation results of the constant speed and constant spacing vehicle platoon are shown in Figures 3-11

[0059] Figures 5-7 The distributed observer of the follower vehicle corresponding to the position information estimation error of the leader, the speed information estimation error and the acceleration information estimation error of the leader are respectively represented by green, red and blue, which correspond to vehicles 1-3. After 10 seconds, the estimation errors of the position, speed and acceleration of the leader of the follower vehicles in the vehicle platoon tend to 0, which shows that the distributed observer designed in the application can still achieve accurate estimation of the state information of the leader in the sparse communication topology.

[0060] Figures 8-10 The black line in the figure represents the disturbance received by the follower vehicles 1-3, and the red line is the estimation value of the disturbance by the extended state observer of each follower. When the vehicle is subjected to the unmodeled disturbance shown by the black line, the extended state observer on the follower vehicle can achieve real-time accurate tracking of the unmodeled disturbance, as shown by the red dashed line.

[0061] Figures 11-13 The position tracking error, speed tracking error and acceleration tracking error of the leader of the follower vehicle under the condition of ensuring the safety interval are respectively represented by green, red and blue, which correspond to vehicles 1-3. After 10 seconds, the position, speed and acceleration tracking errors of the vehicles in the vehicle platoon tend to 0, which shows that the vehicles in the vehicle platoon maintain a constant safety interval and maintain a constant speed. That is, the vehicle platoon realizes safe driving under the control strategy designed in the application.

[0062] Note: In order to facilitate representation, there are two-digit subscripts in the legend. The ten-digit number represents the vehicle number, and the unit digit represents the state (1 represents the position related, 2 represents the speed related, and 3 represents the acceleration related). For example: Figures 3-11 The position error in the observation error of the distributed observer on the first vehicle is represented by The speed error in the tracking error of the third vehicle to the leader is represented by .​

Claims

1. An observer-based intelligent vehicle cluster safety platoon control method, characterized by: The method comprises the following steps: S1: The intelligent vehicle cluster includes 1 leader vehicle and 1 follower vehicle, > 0; for a single follower vehicle, a follower third-order longitudinal dynamics mathematical model is constructed, which contains a three-dimensional state vector of the follower vehicle , , , , The position, speed, and acceleration of the first follower vehicle are represented by , respectively, and time is represented by t; a leader third-order longitudinal dynamics mathematical model is constructed, which contains a three-dimensional state vector of the leader vehicle , , , The position, speed, and acceleration of the leader vehicle are represented by , respectively. S2: constructing an inter-vehicle communication relationship of the intelligent vehicle cluster; S3: Based on inter-vehicle communication relationships, a third-order longitudinal dynamics mathematical model of the follower, a third-order longitudinal dynamics mathematical model of the leader, and the first observation gain matrix. Second observation gain matrix Regarding the first Each follower vehicle constructs a sampling-based distributed observer to obtain the first... A follower vehicle The estimated value ;based on and Construct a distributed observer to estimate the error; S4: Targeting the A follower vehicle, constructing a four-dimensional state vector. , Indicates unmodeled disturbances, based on Construct an extended state observer, and obtain the results through the extended state observer. The estimated value , , , , They are respectively , , , The estimated value; based on and Construct an extended state observer to estimate the error; S5: defining a safety gap between the first follower vehicle and the leader vehicle; constructing a safety distance tracking error of the first follower vehicle to the leader vehicle based on the intelligent vehicle cluster safety formation control strategy and the safety gap ​​​​ S6: constructing a DoS attack model; constructing a segmented distributed observer estimation error under DoS attack based on the DoS attack model and the distributed observer estimation error; constructing a safe distance tracking error under DoS attack based on the segmented distributed observer estimation error, the extended state observer estimation error and the safe distance tracking error; S7: based on the distributed observer estimation error, the extended state observer estimation error, the safety distance tracking error and the DoS attack model, a quadratic Lyapunov function is constructed, and the control gain matrix is obtained by solving the Lyapunov function , the first observation gain matrix and the second observation gain matrix ; S8: realizing the control of the intelligent vehicle cluster by combining the distributed observer, the extended state observer, the DoS attack model and the intelligent vehicle cluster safe formation control strategy.

2. The observer-based intelligent vehicle platoon safety control method of claim 1, wherein the steps of In S1, the follower third-order longitudinal dynamics mathematical model is: （1）； wherein, , , denotes the power system time delay; denotes the control input of the th follower vehicle, denotes the unmodeled disturbance of the th follower vehicle; The leader third-order longitudinal dynamics mathematical model is: （2）； wherein, represents an output value of a mathematical model of the third order longitudinal dynamics of the leader, .

3. The observer-based intelligent vehicle platoon safety control method of claim 2, wherein: In step S2, the inter-vehicle communication is realized through V2V communication, and the inter-vehicle communication relationship includes a communication relationship between follower vehicles and a communication relationship between a leader vehicle and follower vehicles, the communication relationship between follower vehicles is a undirected topological graph wherein, represents a label of a follower vehicle; is an adjacency matrix, , and ; if the first follower vehicle and the first follower vehicle can communicate, then ; otherwise, ; the communication relationship between the leader vehicle and the follower vehicles is if the first follower vehicle can receive information of the leader, then ; otherwise, .

4. The observer-based intelligent vehicle platoon safety control method of claim 3, wherein: In step S3, the distributed observer is: （3）； wherein, denotes the output value of the distributed observer of the i-th following vehicle at the time instant denotes the output value of the distributed observer of the i-th following vehicle at the time instant and denotes the output value of the distributed observer of the i-th following vehicle at the time instant denotes the output value of the distributed observer of the i-th following vehicle at the time instant denotes the output value of the distributed observer of the i-th following vehicle at the time instant denotes the output value of the distributed observer of the i-th following vehicle at the time instant denotes the number of samples, denotes the sampling period;​ The distributed observer estimation error is: （4）。 5. The observer-based intelligent vehicle platoon safety formation control method according to claim 4, characterized in that: In step S4, the extended state observer is: （5）； In conjunction with , formula (4) can be restated as: （6）； wherein , , , represents a zero matrix of 1 row and 3 columns; , represents an observer gain factor, > 0, and when the configuration matrix is chosen , it is necessary to ensure that the real part of the eigenvalues of the configuration matrix is negative. The extended state observer estimation error is: （7）。 6. The observer-based intelligent vehicle platoon safety formation control method according to claim 5, characterized in that: In step S5, the intelligent vehicle cluster safe formation control strategy is: （8）； wherein , , denotes a 3-dimensional identity matrix, denotes a 3x1 zero matrix; , denotes the safety margin; The safe distance tracking error is: （9）。 7. The observer-based intelligent vehicle platoon safety control method of claim 6, wherein: In step S6, the DoS attack model comprises: First sub-interval of non-attack: ; First Sub-attack interval: ; wherein, denotes the time of end of the denotes the time of start of the denotes the duration of the n-th attack; Because attackers have limited resources, DoS attacks can occur in any... Time's up The total duration of all attacks within the time interval and attack frequency Subject to the following conditions: ， ； wherein , , , is a normal number.

8. The observer-based platoon control method for intelligent vehicle cluster safety according to claim 7, characterized in that the step In S6, the segmented distributed observer estimation error is: （10）； wherein , wherein is the incidence matrix, , ; Since the extended state observer is not affected by network attacks, the extended state observer estimation error of all follower vehicles can be expressed as: （11）； wherein , , ; The safe distance tracking error under DoS attack is: （12）； wherein .

9. The observer-based intelligent vehicle platoon safety control method of claim 8, wherein: In step S7, the quadratic Lyapunov function is: ； ； ； ； ； ； wherein , , , , , , , , , is a positive definite matrix, is an integral variable; For Taking the derivative of the distributed observer estimation error, the augmented state observer estimation error and the safety distance tracking error, we can obtain the stability conditions for the global ultimate consensus boundedness under the disturbance of unmodeled disturbances and DoS attacks. The global final consistent bounded stability condition is: ； wherein, , , , , , , are all normal numbers, and the gain matrix , and are obtained simultaneously.

Images