Networked vehicle queue collaborative optimization safety control method and system

Abstract

The invention discloses a networked vehicle queue collaborative optimization safety control method and a networked vehicle queue collaborative optimization safety control system. Firstly, a leading vehicle and following vehicle queue system model is established, and vehicle state information is collected through a sensor; a distributed communication time delay elastic observer is designed to estimate the state of the pilot vehicle for the abnormity possibly existing in the inter-vehicle communication link. And the system makes a difference between the state of the following vehicle and the real state of the pilot vehicle or the state of the observer according to a communication link state judgment result, and an error system is constructed. Then, based on a value function, a Hamiltonian and an HJB equation are derived, and an optimal control law is solved. And finally, an evaluation and execution neural network in reinforcement learning is utilized to respectively approach the value function and the control law, and an optimized optimal control law is obtained through iterative calculation and is output to the following vehicle. According to the method, multi-vehicle-queue collaborative optimization control is realized, and the stability and robustness of the system under hostile attacks and abnormal communication are effectively improved.

Description

Technical Field

[0001] This invention belongs to the field of connected multi-vehicle cooperative platooning control technology, and particularly relates to a method and system for optimized safety control of connected vehicle platooning. Background Technology

[0002] With the rapid development of intelligent transportation systems, multi-vehicle queuing systems, as a core component, have attracted much attention due to their significant advantages in improving road traffic efficiency, safety, and resource utilization. However, the rapid increase in the number of motor vehicles, while bringing convenience to people's travel, has also given rise to a series of social problems, such as frequent traffic accidents, increased road congestion, and environmental pollution. The collaborative control of distributed multi-vehicle queuing systems relies heavily on real-time information interaction between vehicle networks. However, in complex and dynamic network environments, the inherent transmission latency characteristics of connected vehicle communication links and malicious attacks (such as denial-of-service attacks) will severely undermine the real-time performance and consistency necessary for collaborative control of multi-vehicle queuing. Specifically, communication latency usually leads to asynchronous updates of state information between vehicles, which in turn causes a decline in the performance of the queuing controller and even security risks. Denial-of-service attacks, on the other hand, periodically or randomly interrupt the transmission of critical data, directly causing the failure of traditional cooperative adaptive cruise control algorithms based on the assumption of continuous communication. Therefore, in a hybrid model that simultaneously considers communication latency and DoS attacks, designing a controller that combines distributed architecture, attack robustness, and optimal control to ensure the stability, safe spacing, and energy minimization of multi-vehicle platoons under malicious attacks has become a critical scientific problem urgently needing to be solved in the field of cybersecurity and cooperative control of intelligent transportation systems. Currently, research on the cooperative control problem of multi-vehicle platoon systems has attracted widespread attention from scholars in control theory and control engineering, and a series of important research results have been achieved at both the theoretical research and technical implementation levels.

[0003] In the field of longitudinal motion control, multi-vehicle platooning systems, due to their ability to autonomously adjust vehicle throttle opening and braking intensity, increase the automatic control capability of vehicles in the longitudinal direction, thereby reducing traffic accidents caused by driver fatigue due to prolonged vehicle operation. Currently, many control technologies are used in multi-vehicle platooning systems. However, existing control methods still have the following problems:

[0004] 1. Existing research findings are all based on the ideal assumptions of instantaneous, delay-free communication and continuous, stable vehicle-to-vehicle communication links. However, in the actual operation scenarios of multi-vehicle platooning in autonomous driving, sudden situations such as interruptions or delays in vehicle-to-vehicle communication links will affect system stability and even create safety hazards.

[0005] 2. Existing control strategies mostly only consider their control objectives and system stability, but often lead to huge energy consumption. Therefore, how to apply optimal control methods to multi-vehicle platooning systems to minimize energy consumption while ensuring platoon control objectives is the key problem this research aims to solve. Summary of the Invention

[0006] The technical objective of this invention is to address the technical problems of current multi-vehicle queue control methods, such as poor stability due to inter-vehicle communication interruption or communication delay, and high energy consumption, by providing a collaborative optimization safety control method and system for connected vehicle queues.

[0007] To achieve the above technical objectives, the embodiments of this application adopt the following technical solutions.

[0008] In a first aspect, embodiments of this application provide a safety control method for coordinated optimization of connected vehicle queues, including:

[0009] Establish a navigator vehicle system model and a multi-vehicle platoon system model; obtain the status of the following vehicles by combining data collected by sensors on the following vehicles with the multi-vehicle platoon system model; obtain the actual status of the navigator vehicle by combining data collected by sensors on the navigator vehicle with the navigator vehicle system model.

[0010] Based on the aforementioned navigator system model, a distributed communication delay elastic observer is designed. The distributed communication delay elastic observer is used to estimate the navigator state and output the observer state in the case of abnormal inter-vehicle communication link.

[0011] Determine the status of the inter-vehicle communication link. If the communication link is normal, subtract the status of the following vehicle from the actual status of the lead vehicle. If the communication link is abnormal, subtract the status of the following vehicle from the status of the observer to obtain the error system consisting of the error status and the error dynamics equation. Output the error status and the corresponding error system.

[0012] Based on the error state and error system, firstly, the value function of the error system is defined, then the Hamiltonian is constructed based on the value function, and the Hamiltonian is set to 0 to obtain the HJB equation. Subsequently, under the constraints of the HJB equation, the partial derivative of the Hamiltonian with respect to the control input is obtained to obtain the optimal control law, and finally the optimal control law is output.

[0013] Using the optimal control law as input, the value function and the optimal control law are approximated by the evaluation neural network and the execution neural network, respectively, and the optimal control law is obtained by iterative calculation. The optimized optimal control law is then output to the following vehicles in the multi-vehicle queuing system to achieve collaborative optimization control of the multi-vehicle queuing.

[0014] Secondly, embodiments of this application provide a connected vehicle platoon collaborative optimization safety control system, including:

[0015] The vehicle platooning system is used to establish a lead vehicle system model and a follower vehicle platooning system model; by collecting data from sensors on the following vehicles and combining it with the follower vehicle platooning system model, the status of the following vehicles is obtained; by collecting data from sensors on the lead vehicle and combining it with the lead vehicle system model, the actual status of the lead vehicle is obtained.

[0016] The distributed observer module is used to design a distributed communication delay elastic observer based on the lead vehicle system model. The distributed communication delay elastic observer is used to estimate the state of the lead vehicle in the case of abnormal inter-vehicle communication link and output the observer state.

[0017] The system transformation module is used to determine the status of the inter-vehicle communication link. If the communication link is normal, the following vehicle status is subtracted from the actual status of the lead vehicle. If the communication link is abnormal, the following vehicle status is subtracted from the observer status to obtain the error system composed of the error status and the error dynamics equation, and the error status and the corresponding error system are output.

[0018] The optimal control law calculation module is used to first define the value function of the error system based on the error state and the error system, then construct the Hamiltonian based on the value function, derive the HJB equation by setting the value of the Hamiltonian to 0, and then, under the constraint of satisfying the HJB equation, calculate the partial derivative of the Hamiltonian with respect to the control input to obtain the optimal control law and output the optimal control law.

[0019] The reinforcement learning module takes the optimal control law as input, and uses a judgment neural network and an execution neural network to approximate the value function and the optimal control law respectively. Iterative calculation is performed to obtain the optimized optimal control law, and the optimized optimal control law is output to the following vehicles in the multi-vehicle queuing system to realize the collaborative optimization control of the multi-vehicle queuing.

[0020] Compared with existing technologies, the connected vehicle queue collaborative optimization safety control method and system provided by this invention have the following beneficial effects:

[0021] First, for a type of multi-vehicle queuing system with abnormal convoy communication links (such as communication delays and communication connection drops caused by DoS attacks), this invention designs an elastic distributed navigator state observer resistant to communication abnormalities, which effectively improves the stability and robustness of the control process.

[0022] Secondly, this invention applies the concept of optimal control, which can minimize the cost of the system while ensuring system stability, thus making it more applicable than previous studies.

[0023] Third, this invention uses the idea of ​​reinforcement learning, applying a judgment neural network and an execution neural network to approximate the value function and control law of the system respectively, which can achieve the control objective under the premise of unknown system dynamics.

[0024] It should be understood that the summary section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0025] The accompanying drawings described herein are for illustrative purposes only and are not intended to limit the scope of this application in any way. Furthermore, the shapes and scales of the components in the drawings are merely illustrative to aid in understanding this application and do not specifically limit the shapes and scales of the components. Those skilled in the art, guided by the teachings of this application, can select various possible shapes and scales to implement this application according to specific circumstances. In the drawings:

[0026] Figure 1 This is a schematic diagram of the safety control system framework for the connected vehicle queue collaborative optimization in the embodiment.

[0027] Figure 2 These are effect diagrams of a distributed communication latency elastic observer applicable to communication latency and DoS attacks in the embodiments; where (a) is the position effect diagram, (b) is the velocity effect diagram, and (c) is the acceleration effect diagram.

[0028] Figure 3 This is a schematic diagram showing the positions of the lead vehicle and the follower vehicle in the embodiment.

[0029] Figure 4 This is a speed diagram of the lead vehicle and the following vehicle in the embodiment.

[0030] Figure 5 This is a schematic diagram of the acceleration of the lead vehicle and the following vehicle in the embodiment.

[0031] Figure 6 This is the first example in the embodiment. Control law of following vehicle Renderings. Detailed Implementation

[0032] To enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this application.

[0033] The terms "step one," "step two," "step three," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance, limiting the order of steps, or implicitly specifying the number of technical features indicated.

[0034] In response to the anomaly of the convoy communication link, this invention constructs an adaptive distributed elastic state observer to estimate the state of the lead vehicle, designs an optimal control strategy, and constructs a distributed cooperative optimization queue controller that is resistant to communication delay and DoS attacks. This achieves the dual control objectives of global Nash equilibrium convergence and energy consumption minimization for each following vehicle.

[0035] Example 1: A safety control method for collaborative optimization of connected vehicle queues, including:

[0036] Step 1: Establish the lead vehicle system model and the following multi-vehicle platoon system model; obtain the following vehicle status by combining the data collected by the sensors on the following vehicles with the following multi-vehicle platoon system model; obtain the lead vehicle's actual status by combining the data collected by the sensors on the lead vehicle with the lead vehicle system model.

[0037] Step 2: Based on the lead vehicle system model, design a distributed communication delay elastic observer. The distributed communication delay elastic observer is used to estimate the state of the lead vehicle in the case of abnormal inter-vehicle communication link and output the observer state.

[0038] Step 3: Determine the status of the inter-vehicle communication link. If the communication link is normal, subtract the following vehicle's status from the lead vehicle's actual status. If the communication link is abnormal, subtract the following vehicle's status from the observer's status to obtain the error system consisting of the error status and the error dynamics equation. Output the error status and the corresponding error system.

[0039] Step 4: Based on the error state and error system, first define the value function of the error system, then construct the Hamiltonian based on the value function, set the value of the Hamiltonian to 0 to obtain the HJB equation, and then, under the constraints of the HJB equation, take the partial derivative of the Hamiltonian with respect to the control input to obtain the optimal control law, and output the optimal control law.

[0040] Step 5: Using the optimal control law as input, the evaluation neural network and the execution neural network are used to approximate the value function and the optimal control law respectively. The optimized optimal control law is obtained through iterative calculation. The optimized optimal control law is then output to the following vehicles in the multi-vehicle queuing system to achieve collaborative optimization control of the multi-vehicle queuing.

[0041] In this embodiment, the navigator system model established in step one generates the following equations:

[0042] (1)

[0043] (2)

[0044] Among the variables It's the navigator vehicle's status. It is the output of the navigator. and These are the system matrix and output matrix of the navigator system, respectively.

[0045] In this embodiment, the multi-vehicle queuing system model is described by the following differential equation:

[0046] (3)

[0047] Among them, intermediate variables are defined. , and The first The following vehicle Displacement, velocity, and acceleration at time t, continuous function and These are the drift state and input gain of the following vehicle, respectively. It is the system number Time of the first Control law for a following vehicle, intermediate variables Where air density, vehicle cross-sectional area, and air drag coefficient are respectively represented by... , and express, Represents engine constants. ,in Represents rolling resistance. Represents gravity. The rolling resistance coefficient, It is gravitational acceleration. It's the road slope. For the first The quality of the following vehicles.

[0048] The main function of the distributed communication delay resilience observer is to estimate the state of the lead vehicle when communication links in a multi-vehicle queuing system are abnormal (such as communication delay and DoS attack).

[0049] In this embodiment, the distributed communication delay elastic observer takes the following form:

[0050] (4)

[0051] (5)

[0052] in, Representing the The status of the lead vehicle as observed by the following vehicles. Representing the The observer gain corresponding to the vehicle following the vehicle. It exists in the first Communication latency between the following vehicle and the lead vehicle. It exists in the first Following the car and the first Communication latency of the following vehicle It is the first The safety status between the following vehicle and the lead vehicle. It is the first The safety status between the following vehicle and the lead vehicle. It is the lead car and the first Adjacent elements between following vehicles It is the first The following vehicle and the first Adjacent elements between following vehicles The total number of following vehicles. Representing the The status of the lead vehicle as observed by the following vehicles. Indicates the current time. This indicates the status of the lead vehicle affected by communication delays. It is the first The signal output by the lead vehicle observed by the following vehicles. C is the system matrix of the navigator system, and C is the output matrix of the navigator system.

[0053] The main purpose of step three is to obtain the corresponding error state and error dynamic equation by subtracting the state of the following vehicle system from the state of the lead vehicle under the premise of a safe state. The system can then be transformed into an error equation and a controller can be designed using the optimal control method.

[0054] By subtracting the model of the multi-vehicle platoon system (Equation (3)) from the model of the lead vehicle system that generates the reference signal (Equation (1)), we can obtain the following results.

[0055] (6)

[0056] (7)

[0057] in Indicates the first The derivative of the following vehicle's state. , , These represent the navigation vehicle system matrix. Lines 1, 2, and 3 , , For the first Safety distance vector between following vehicle and lead vehicle The first, second, and third components, Representing the The state error between the following vehicle and the lead vehicle. For the drift state of the error system, Input gain to the error system, For the first Control law of following vehicle , It's the status of the lead car.

[0058] The main function of step four is to solve for the value function, HJB equation and control law corresponding to the error system based on the error system obtained from the system transformation module, and output the results to the reinforcement learning module for approximation.

[0059] In the embodiment, the first step is to give the first... The value function of the error system of the following vehicle is of the following form:

[0060] (8)

[0061] in For utility function, For the first The error status of the following vehicle. For the first The control law of the following vehicle For the first The control law of the following vehicle The time variable is used for integration. Function ,in , and All are positive definite matrices.

[0062] Its corresponding Hamiltonian can be expressed as

[0063] (9)

[0064] in, Indicates all related to the first The following vehicles communicate with each other.

[0065] Taking the Hamiltonian obtained in formula (9) as 0, the HJB equation corresponding to the error system (7) can be obtained as follows:

[0066] (10)

[0067] Subsequently, the Hamiltonian shown in (9) is compared with the first... Control law (control input) for a following vehicle Taking the partial derivative, we can obtain the optimal control law applicable to the error system (7), which has the following form:

[0068] (11)

[0069] The main function of step five is to approximate the control law and value function of the error system (7) by executing the neural network and evaluating the neural network respectively under the premise of unknown dynamics of the multi-vehicle following system (3), and to iteratively calculate the optimal control law based on the approximation result, and finally output the result to the following vehicle in the multi-vehicle queuing system.

[0070] Among them, the Value function of the following vehicle and the Control law of following vehicle The approximation can be performed using the evaluation neural network and the execution neural network respectively, in the following forms:

[0071] (12)

[0072] (13)

[0073] In the above formulas (12) and (13), and These represent the weights of the evaluation neural network and the execution neural network, respectively. and These represent the basis functions for evaluating the neural network and executing the neural network, respectively. and These represent the approximation errors of the evaluation neural network and the execution neural network for the value function and the control law, respectively.

[0074] Example 2: Based on the same inventive concept as the connected vehicle queue collaborative optimization safety control method provided in the above examples, this application also provides a connected vehicle queue collaborative optimization safety control system, including a vehicle queue system, a distributed observer module, a system transformation module, an optimal control law calculation module, and a reinforcement learning module.

[0075] The vehicle platooning system is used to establish a lead vehicle system model and a follower vehicle platooning system model; by collecting data from sensors on the following vehicles and combining it with the follower vehicle platooning system model, the status of the following vehicles is obtained; by collecting data from sensors on the lead vehicle and combining it with the lead vehicle system model, the actual status of the lead vehicle is obtained.

[0076] The distributed observer module is used to design a distributed communication delay elastic observer based on the lead vehicle system model. The distributed communication delay elastic observer is used to estimate the state of the lead vehicle in the case of abnormal inter-vehicle communication link and output the observer state.

[0077] The system transformation module is used to determine the status of the inter-vehicle communication link. If the communication link is normal, the following vehicle status is subtracted from the actual status of the lead vehicle. If the communication link is abnormal, the following vehicle status is subtracted from the observer status to obtain the error system consisting of the error status and the error dynamics equation. The error status and the corresponding error system are then output.

[0078] The optimal control law calculation module is used to first define the value function of the error system based on the error state and the error system, then construct the Hamiltonian based on the value function, derive the HJB equation by setting the value of the Hamiltonian to 0, and then, under the constraint of satisfying the HJB equation, calculate the partial derivative of the Hamiltonian with respect to the control input to obtain the optimal control law and output the optimal control law.

[0079] The reinforcement learning module takes the optimal control law as input, and uses a judgment neural network and an execution neural network to approximate the value function and the optimal control law respectively. Iterative calculations yield the optimized optimal control law, which is then output to the following vehicles in the multi-vehicle queuing system to achieve collaborative optimization control of the multi-vehicle queuing.

[0080] In the embodiment, the optimal control law obtained by solving the HJB equation can ensure that the multi-vehicle queuing system achieves the effect of minimizing energy consumption or cost while realizing the cooperative control objective.

[0081] In some embodiments, the optimal controller design structure for a type of multi-vehicle queuing system resistant to communication latency and DoS attacks is as follows: Figure 1 As shown. In the multi-vehicle queuing system transformation module, the status of the following vehicle... By the status of the lead car and the state of the distributed communication delay elastic observer in the distributed observer module The difference is used to obtain the corresponding error. The error dynamics equation is then used, and the error is input into the optimal control law calculation module to obtain the corresponding value function. HJB equations and optimal control laws The optimal control law calculated by the optimal control law calculation module is used as the input to the reinforcement learning module. In the reinforcement learning module, a judgment neural network is designed. and execution neural network Approximate the optimal value function under the premise of unknown following vehicle dynamics equations. and optimal control law The optimal control law is then output to each following vehicle in the multi-vehicle platooning system. The design objective of this invention is to further prevent communication delays and DoS attacks that could cause communication disconnections in real-world control scenarios, thereby effectively improving the system's stability and robustness, while incorporating optimal control.

[0082] Simulation results are as follows Figures 2-6 As shown. Figure 2 This diagram illustrates the effect of a distributed communication latency resilience observer applicable to communication latency and DoS attacks. The observer state shown in the diagram can effectively estimate the navigator state. Figure 3 This is a diagram showing the positions of the lead vehicle and the following vehicles. Figure 4 This is a speed diagram of the lead vehicle and the following vehicle. Figure 5 This is a diagram illustrating the acceleration of the lead vehicle and the following vehicles. Figure 2 The results show that the proposed distributed communication delay elastic observer can effectively estimate the state of the navigator vehicle under conditions of communication delay and DoS attacks. Based on... Figure 3 It can be verified that the following vehicle can stably follow the lead vehicle while maintaining a safe distance. Figure 4-5 It can be concluded that the following vehicle can keep up with the speed and acceleration of the lead vehicle in a relatively short period of time. Figure 6 The diagram shows the effect of the control signals, demonstrating that the control signals are reasonable during the process. The simulation results above verify that the following vehicles in the controlled multi-vehicle platoon system can achieve good following performance from the lead vehicle under the optimal controller proposed in this invention.

[0083] This invention is not limited to this embodiment. Any equivalent concept or modification within the technical scope disclosed in this invention shall be included within the protection scope of this invention.

[0084] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or physical entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer can be, for example, a personal computer, a laptop computer, or a tablet computer, or any combination of these devices.

[0085] The above provides a detailed description of the connected vehicle queue collaborative optimization safety control method and system provided in this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the concept of this application and should not be construed as limiting the scope of protection of this application.

Claims

1. A safety control method for collaborative optimization of connected vehicle platoons, characterized in that, include: Establish a navigator vehicle system model and a multi-vehicle platoon system model; obtain the status of the following vehicles by combining data collected by sensors on the following vehicles with the multi-vehicle platoon system model; obtain the actual status of the navigator vehicle by combining data collected by sensors on the navigator vehicle with the navigator vehicle system model. Based on the aforementioned navigator system model, a distributed communication delay elastic observer is designed. The distributed communication delay elastic observer is used to estimate the navigator state and output the observer state in the case of abnormal inter-vehicle communication link. Determine the status of the inter-vehicle communication link. If the communication link is normal, subtract the status of the following vehicle from the actual status of the lead vehicle. If the communication link is abnormal, subtract the status of the following vehicle from the status of the observer to obtain the error system consisting of the error status and the error dynamics equation. Output the error status and the corresponding error system. Based on the error state and error system, firstly, the value function of the error system is defined, then the Hamiltonian is constructed based on the value function, and the HJB equation is derived by setting the value of the Hamiltonian to 0. Subsequently, under the constraint of satisfying the HJB equation, the partial derivative of the Hamiltonian with respect to the control input is obtained to obtain the optimal control law, and the optimal control law is output. Using the optimal control law as input, the value function and the optimal control law are approximated by a judgment neural network and an execution neural network, respectively. The optimized optimal control law is obtained by iterative calculation, and the optimized optimal control law is output to the following vehicles in the multi-vehicle queuing system to realize the collaborative optimization control of the multi-vehicle queuing.

2. The connected vehicle queue collaborative optimization safety control method according to claim 1, characterized in that, The abnormalities in the inter-vehicle communication link include two scenarios: communication delay and DoS attack.

3. The connected vehicle queue collaborative optimization safety control method according to claim 1, characterized in that, The navigator system model is generated by the following equations: ； ； Among the variables It's the navigator vehicle's status. It is the output of the navigator. and These are the system matrix and output matrix of the navigator system, respectively.

4. The connected vehicle queue collaborative optimization safety control method according to claim 1, characterized in that, The model of the multi-vehicle queuing system is described by the following differential equation: ； Among them, intermediate variables are defined. , and The first The following vehicle Displacement, velocity, and acceleration at time t, continuous function and These are the drift state and input gain of the following vehicle, respectively. It is the system number Time of the first Control law for a following vehicle, intermediate variables Where air density, vehicle cross-sectional area, and air drag coefficient are respectively represented by... , and express, Represents engine constants. ,in Represents rolling resistance. Represents gravity. The rolling resistance coefficient, It is gravitational acceleration. It's the road slope. For the first The quality of the following vehicles.

5. The connected vehicle queue collaborative optimization safety control method according to claim 1, characterized in that, The distributed communication delay elastic observer takes the following form: ； ； in, Representing the The status of the lead vehicle as observed by the following vehicles. Representing the The observer gain corresponding to the vehicle following the vehicle. It exists in the first Communication latency between the following vehicle and the lead vehicle. It exists in the first The following vehicle and the first Communication delay between following vehicles It is the first The safety status between the following vehicle and the lead vehicle. It is the first The safety status between the following vehicle and the lead vehicle. It is the lead car and the first Adjacent elements between following vehicles It is the first The following vehicle and the first Adjacent elements between following vehicles The total number of following vehicles. Representing the The status of the lead vehicle as observed by the following vehicles. Indicates the current time. This indicates the status of the lead vehicle affected by communication delays. It is the first The signal output by the lead vehicle observed by the following vehicles. C is the system matrix of the navigator system, and C is the output matrix of the navigator system.

6. A connected vehicle platoon collaborative optimization safety control system, characterized in that, include: The vehicle platooning system is used to establish a lead vehicle system model and a follower vehicle platooning system model; by collecting data from sensors on the following vehicles and combining it with the follower vehicle platooning system model, the status of the following vehicles is obtained; by collecting data from sensors on the lead vehicle and combining it with the lead vehicle system model, the actual status of the lead vehicle is obtained. The distributed observer module is used to design a distributed communication delay elastic observer based on the lead vehicle system model. The distributed communication delay elastic observer is used to estimate the state of the lead vehicle in the case of abnormal inter-vehicle communication link and output the observer state. The system transformation module is used to determine the status of the inter-vehicle communication link. If the communication link is normal, the following vehicle status is subtracted from the actual status of the lead vehicle. If the communication link is abnormal, the following vehicle status is subtracted from the observer status to obtain the error system composed of the error status and the error dynamics equation, and the error status and the corresponding error system are output. The optimal control law calculation module is used to first define the value function of the error system based on the error state and the error system, then construct the Hamiltonian based on the value function, derive the HJB equation by setting the value of the Hamiltonian to 0, and then, under the constraint of satisfying the HJB equation, calculate the partial derivative of the Hamiltonian with respect to the control input to obtain the optimal control law and output the optimal control law. The reinforcement learning module takes the optimal control law as input, uses a judgment neural network and an execution neural network to approximate the value function and the optimal control law respectively, and iteratively calculates the optimized optimal control law. The optimized optimal control law is then output to the following vehicles in the multi-vehicle queuing system to achieve collaborative optimization control of the multi-vehicle queuing.

Images