Networked vehicle formation automatic control method based on anti-noise gradient neurodynamics

Abstract

The invention discloses a networked vehicle formation automatic control method based on anti-noise gradient neurodynamics, belongs to the technical field of vehicle control, and provides a double-integral anti-noise gradient neurodynamics model DINRGND model which is embedded into a vehicle formation controller to enhance the information transmission capability between vehicles, so that the vehicle formation automatic control is realized. Therefore, the synchronism and the stability between the vehicles are ensured; the invention further designs a state estimation method based on the distributed observer, and the precision of state estimation is remarkably improved. Compared with the prior art, the method has the advantages that higher stability and robustness are achieved under noise interference, the problem that a traditional vehicle formation control method is difficult to synchronize in a complex communication environment is solved, good stability and robustness among vehicles can be kept under the noise interference condition, road safety is effectively improved, and the method is suitable for popularization and application. The method provides a solid theoretical basis for automatic control of unmanned or networked vehicle formations, and has a wide application prospect.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of vehicle control, and particularly relates to a networking vehicle platoon automatic control method based on anti-noise gradient neural dynamics. BACKGROUND

[0002] With the rapid development of intelligent transportation systems and automatic driving technologies, vehicle platoon control has gradually become an important means to improve road traffic efficiency, reduce energy consumption, and ensure traffic safety. Vehicles travel in a tandem queue in the same lane, and their spacing is precisely controlled by an automatic control system. With the help of vehicle-to-vehicle communication (V2V), vehicles can expand their detection range and obtain key data of surrounding vehicles, such as position, speed, and acceleration. By exchanging key data between vehicle platoons in real time, efficient coordination between vehicles can be achieved, thereby effectively alleviating traffic congestion and improving road resource utilization. Especially in the environment of unmanned and intelligent networking vehicles, platoon automatic control shows greater application potential.

[0003] For actual application scenarios, vehicle platoon control often faces the interference of complex traffic environments. Environmental noise can cause errors in the perception signals between vehicles, leading to a decline in control performance. These problems can easily lead to unstable distances between vehicles and even trigger rear-end collision accidents, thereby posing a serious threat to the safety and stability of vehicle platoon operation. Existing platoon control methods mostly use traditional controller theories such as consensus and distributed control. These methods can achieve vehicle queue driving under ideal conditions, but when external disturbances such as communication noise, sensor errors, etc. affect the received communication signals, the control performance is significantly reduced, which can easily lead to queue oscillation. Currently, there is a lack of an effective control method that can maintain stable platoon in a multi-source noise environment. Therefore, there is an urgent need to propose a vehicle platoon control method that can effectively suppress noise interference and improve the robustness and convergence of the automatic control system to ensure the stable and safe coordinated operation of networking vehicles in actual road environments.

[0004] In recent years, with the development of Internet of Vehicles technology, deep learning has shown broad application prospects in vehicle cooperative control and platoon formation. Compared with traditional control methods, deep learning can model and optimize complex traffic environments through an end-to-end approach, with the advantages of adaptive learning of traffic dynamics and real-time processing of multi-source data. A large number of researchers have attempted to introduce convolutional neural networks (CNN) and reinforcement learning (DRL) into vehicle platoon control for vehicle distance maintenance and traffic flow prediction. In addition, neural dynamics (ND) models with parallel processing capabilities have shown excellent performance in computing power and data processing capabilities and have been applied to vehicle platoon control. For example, a distributed control method based on time-varying parameter ZND can effectively maintain vehicle platoon stability in a V2V communication environment and has fast response and anti-interference ability to periodic noise. However, the types of noise in real traffic environments are more complex and diverse, and how to use ND models with wide-ranging interference suppression capabilities to achieve robust platoon control of multi-vehicle systems still faces the following challenges:

[0005] (1) Noise sensitivity. Traditional control methods usually ignore the delay and interference of vehicle-to-vehicle (V2V) signal transmission in noisy environments. And most methods often assume that the noise is single, lack of theoretical stability guarantee under multi-source noise, and it is usually difficult to prove the stability of the platoon under multi-source noise. When there is multi-source noise, the error between vehicles is easy to accumulate, causing the formation to deviate and even collision risk.

[0006] (2) Existing control methods can achieve vehicle platoon in weak noise interference, but in strong noise environment, state information transmission is easy to distort, making control instructions inaccurate or not timely, causing vehicle oscillation. It cannot guarantee the stability of the distance between vehicles, and there is a serious road safety hazard.

[0007] (3) Existing deep learning methods often need to rely on a large amount of training data, such as vehicle trajectories and V2V communication information. And it is difficult to obtain data covering various weather, roads, and emergency situations in real traffic, and when the data is insufficient, it is easy to cause overfitting. At the same time, such methods have high computational complexity and are difficult to deploy in limited vehicle-mounted devices, and lack robustness to environmental disturbances. At the same time, traditional observer methods also have limitations, and their performance is heavily dependent on parameter settings, and when the environment changes beyond the set range, it is often difficult to guarantee stability and reliability. SUMMARY

[0008] In view of the above problems in the prior art, the networking vehicle platoon automatic control method based on anti-noise gradient neural dynamics provided by the present application solves the problem that the method does not consider that the vehicle platoon control is easily disturbed by signal interference of environmental noise and system uncertainty, signal synchronization is difficult in a complex communication environment, vehicle spacing is unstable, even a rear-end collision accident occurs, and the safety and stability of the vehicle fleet operation are seriously affected, and the robustness and convergence of the vehicle platoon automatic control under multi-source noise disturbance are enhanced by proposing a gradient neural dynamics model (DINRGND) with anti-noise capability.

[0009] In order to achieve the above-mentioned purpose of the application, the technical scheme adopted by the present application is as follows: the networking vehicle platoon automatic control method based on anti-noise gradient neural dynamics comprises:

[0010] a vehicle longitudinal dynamics model is constructed;

[0011] a communication topology structure of the networking vehicle is constructed, and a virtual leader vehicle is introduced into the communication topology structure to provide a reference trajectory;

[0012] a distributed observer and a controller are constructed for the vehicle longitudinal dynamics model;

[0013] a double-integral anti-noise gradient neural dynamics model is constructed and embedded into the controller;

[0014] In the vehicle platoon control process, the vehicle is controlled by the controller based on the constructed communication topology structure under error attenuation and noise suppression, and the state estimation result of the virtual leader vehicle is dynamically corrected by the integral compensation mechanism of the distributed observer using historical error information, so as to realize the networking vehicle platoon automatic control.

[0015] Further, the vehicle longitudinal dynamics model is a dynamic model introducing external disturbance, which is expressed as:

[0016] ;

[0017] In the formula, denotes the time derivative of the position of the vehicle , denotes the time derivative of the speed of the vehicle , denotes the time derivative of the acceleration of the vehicle , denotes the speed of the vehicle , denotes the acceleration of the vehicle , denotes the controller input of the linearized vehicle , denotes the noise disturbance suffered by the vehicle , wherein, , , This represents the new input after feedback linearization. This represents disturbances and model uncertainties. This represents the inertial delay in the longitudinal dynamics of a vehicle.

[0018] Furthermore, in the communication topology of the connected vehicle In the middle, node Indicates vehicle, side This indicates the information transmission relationship between vehicles. Represents the communication topology The adjacency matrix, Indicates vehicle To the vehicle Connection weight for transmitting information, if vehicle With vehicles If there is a connection, then =1, otherwise =0; where, , , Indicates the total number of vehicles;

[0019] In the communication topology Through the Laplace matrix Describes the communication connection status between vehicles; where the elements in the Laplace matrix... Represented as:

[0020] ;

[0021] In the formula, Indicates vehicle To the vehicle Connection weight for transmitting information, ;

[0022] In the communication topology In, through a diagonal matrix Describes the communication connection status between the introduced virtual pilot vehicle and other vehicles, when the vehicle When the virtual navigator vehicle can directly receive information from the virtual navigator vehicle, the virtual navigator vehicle sends information to the vehicle. Connection weight for transmitting information ,otherwise .

[0023] Furthermore, vehicles The distributed observer is represented as:

[0024] ;

[0025] In the formula, representing a vehicle an estimate of a virtual lead vehicle position, representing a vehicle an estimate of a virtual lead vehicle position, representing a vehicle an estimate of a virtual lead vehicle position, representing a virtual lead vehicle position, representing a vehicle to a vehicle a connection weight transmitting information, representing a virtual lead vehicle to a vehicle a connection weight transmitting information, , and representing first, second and third positive fixed convergence parameters, respectively, representing a Jacobian matrix, representing a current integration upper limit time, representing a virtual time variable during integration, , representing a total number of vehicles;

[0026] representing a vehicle a derivative of an estimate of a virtual lead vehicle speed, representing a vehicle an estimate of a virtual lead vehicle speed, representing a vehicle an estimate of a virtual lead vehicle speed, representing a virtual lead vehicle speed;

[0027] representing a vehicle a derivative of an estimate of a virtual lead vehicle acceleration, representing a vehicle an estimate of a virtual lead vehicle acceleration, representing a vehicle an estimate of a virtual lead vehicle acceleration, representing a virtual lead vehicle acceleration.

[0028] Further, a vehicle controller is represented as:

[0029] ;

[0030] ;

[0031] ;

[0032] wherein a position error of the vehicle , a position error of the vehicle , a speed error of the vehicle , an acceleration error of the vehicle , a nonlinear variable configured as the vehicle , a nonlinear variable configured as the vehicle , , and respectively represent first, second and third positive fixed parameters;

[0033] a position of the vehicle , an expected distance between the vehicle and a virtual lead vehicle, a speed of the vehicle , an acceleration of the vehicle , a difference between an acceleration of the vehicle and an acceleration of the virtual lead vehicle.

[0034] Further, a double-integral anti-noise gradient neural dynamics model is constructed, comprising:

[0035] a vehicle platoon control problem is converted into a nonlinear problem, and an error function corresponding thereto is constructed;

[0036] based on the constructed error function, an auxiliary variable for enhancing system convergence and suppressing noise steady-state error is constructed;

[0037] the constructed auxiliary variable is differentiated, and the auxiliary variable and the differentiated auxiliary variable are substituted into a conditional formula, to obtain the constructed double-integral anti-noise gradient neural dynamics model.

[0038] Further, the auxiliary variable for enhancing system convergence and suppressing noise steady-state error is represented as:

[0039] ;

[0040] the conditional formula is:

[0041] ;

[0042] wherein, the error function is represented as: , and denote the first, second and third positive fixed convergence parameters, respectively, denotes the Jacobian matrix function, denotes the current integral upper limit time, denotes a virtual time variable during integration, denotes an auxiliary variable after derivation.

[0043] Further, the double-integral noise-robust gradient neural dynamics model is expressed as:

[0044] ;

[0045] wherein, denotes the time derivative of the error function .

[0046] The method of the present application is based on the traditional error feedback control, takes the speed, position and acceleration errors as inputs, constructs a double-integral noise-robust gradient neural dynamics model, realizes the rapid decay of errors and the suppression of external noise, and at the same time, estimates the state of the vehicle through the design of an observer, so that the stability and robustness of the automatic control of the vehicle formation can be maintained even under the interference of multiple source noises. Finally, simulation verification is carried out under the conditions of undirected communication topology and directed communication topology. Compared with the prior art, the present application has the following beneficial effects:

[0047] (1) The traditional gradient neural dynamics model is often sensitive to noise and lacks robustness when dealing with the automatic control of networked vehicle formation. The present application proposes a double-integral noise-robust gradient neural dynamics model DINRGND, which introduces a double-integral term based on the traditional gradient neural dynamics model. This model uses gradient for calculation, has lower calculation complexity, and can significantly improve resource utilization efficiency. On this basis, by embedding the proposed DINRGND model into the vehicle formation controller, the controller can actively suppress noise, ensure the synchronous convergence and stable distance between the following vehicles, effectively avoid the risk of rear-end collision, and improve the safety and robustness of the system.

[0048] (2) For some vehicle states that are difficult to obtain directly, such as the acceleration of the following vehicle, the present application designs a new state estimation method based on a distributed observer. The observer increases an integral compensation mechanism in the traditional observer structure, dynamically corrects the estimation result using historical error information, not only improves the accuracy of state estimation, but also enhances the anti-interference ability of the entire vehicle formation automatic control system in a strong noise environment.

[0049] (3) The double-integral anti-noise gradient neural dynamics model DINRGND provided by the application significantly reduces the computational complexity while maintaining stability, facilitating deployment on resource-constrained platforms such as vehicle-mounted controllers, and improving the actual usability of the system. Unlike deep learning methods that rely on a large amount of training data, this model can be directly constructed and solved without learning, thereby avoiding data dependence and overfitting problems. BRIEF DESCRIPTION OF DRAWINGS

[0050] Figure 1 The anti-noise gradient neural dynamics-based networking vehicle platoon automatic control method flowchart provided by the application.

[0051] Figure 2 The undirected communication topology structure diagram provided by the application.

[0052] Figure 3 The vehicle platoon automatic control diagram of the networking vehicle under undirected communication provided by the application; wherein (a) the acceleration of the vehicle; (b) the speed of the vehicle; (c) the position of the vehicle.

[0053] Figure 4 The relationship information diagram between the vehicles and the virtual lead vehicle under undirected communication provided by the application; wherein (a) the distance between the front and rear vehicles; (b) the distance between each vehicle and the virtual lead vehicle; (c) the speed difference between each vehicle and the virtual lead vehicle; (d) the acceleration difference between each vehicle and the virtual lead vehicle.

[0054] Figure 5 The directed communication topology structure diagram provided by the application.

[0055] Figure 6 The vehicle platoon automatic control diagram of the networking vehicle under five-way communication provided by the application; wherein (a) the acceleration of the vehicle; (b) the speed of the vehicle; (c) the position of the vehicle.

[0056] Figure 7 The relationship information diagram between the vehicles and the virtual lead vehicle under directed communication provided by the application; wherein (a) the distance between the front and rear vehicles; (b) the distance between each vehicle and the virtual lead vehicle; (c) the speed difference between each vehicle and the virtual lead vehicle; (d) the acceleration difference between each vehicle and the virtual lead vehicle. DETAILED DESCRIPTION

[0057] The specific embodiments of the application are described below to facilitate understanding of the application by those skilled in the art, but it should be clear that the application is not limited to the scope of the specific embodiments, and for those skilled in the art, it is obvious that various changes are within the spirit and scope of the application as defined and determined by the appended claims, and all applications utilizing the concept of the application are within the scope of protection.

[0058] This invention provides an automatic control method for networked vehicle platooning based on noise-resistant gradient neurodynamics;

[0059] See Figure 1 ,include:

[0060] Construct a longitudinal dynamics model of the vehicle;

[0061] Construct a communication topology for connected vehicles and introduce a virtual navigator vehicle into the communication topology to provide a reference trajectory;

[0062] For the longitudinal dynamics model of the vehicle, a distributed observer and controller are constructed.

[0063] A double-integral noise-resistant gradient neurodynamic model was constructed and embedded into the controller;

[0064] During vehicle platooning control, based on the constructed communication topology, the controller performs vehicle control under error attenuation and noise suppression. Through the integral compensation mechanism of the distributed observer, the state estimation results of the virtual navigator are dynamically corrected using historical error information, thereby realizing automatic control of networked vehicle platooning.

[0065] In this embodiment of the invention, during the construction of the vehicle longitudinal dynamics model, to make the longitudinal dynamics modeling and control of platooned vehicles more feasible, it is necessary to simplify the originally complex dynamic characteristics. This embodiment makes the following assumptions during modeling: negligible tire longitudinal slippage is ignored; the power system is modeled as a first-order function; the vehicle body is assumed to be a rigid and symmetrical structure; the driving and braking torques are considered controllable; and the effects of pitch and yaw motions are ignored. Based on these assumptions, the vehicle longitudinal dynamics model is expressed as follows:

[0066] ;

[0067] In the formula, and Representing vehicles Position and velocity, Indicates vehicle mass. Indicates the mechanical efficiency of a rotating system. Indicates the actual driving / dwelling torque. , and These represent the time derivatives with respect to position, velocity, and actual drive / dwell torque, respectively. Indicates the radius of the tire. This represents the combined air drag coefficient. Represents gravitational acceleration. Indicates the tire rolling resistance coefficient. inertia delay of vehicle longitudinal dynamics, and denotes the desired driving / retaining torque.

[0068] Then, the following feedback linearization technique is adopted:

[0069] ;

[0070] where denotes the new input after linearization.

[0071] For the vehicle , , the following linear model of vehicle longitudinal dynamics is obtained as follows:

[0072] ;

[0073] where denotes the acceleration of the vehicle , denotes the time derivative of the acceleration of the vehicle .

[0074] Since external disturbances are inevitable, external disturbances are introduced, and the dynamic model of the disturbance is represented as:

[0075] ;

[0076] where denotes the disturbance and model uncertainty. Let , the above formula is rewritten as the final vehicle longitudinal dynamics model represented as:

[0077] ;

[0078] where denotes the time derivative of the position of the vehicle , denotes the time derivative of the speed of the vehicle , denotes the time derivative of the acceleration of the vehicle , denotes the speed of the vehicle , denotes the acceleration of the vehicle , denotes the controller input of the vehicle after linearization, denotes the noise disturbance received by the vehicle , where , , denotes the new input after feedback linearization processing, represents disturbance and model uncertainty, represents inertia delay of vehicle longitudinal dynamics.

[0079] In vehicle platoon control, in order to avoid rear-end collision between vehicles, the controller must be reasonably designed. In order to achieve this goal, a virtual leader vehicle is often used as a reference, and its dynamic model can be regarded as an external system generating a desired command trajectory, and its expression is:

[0080] ;

[0081] In the formula, , and respectively represent the position, speed and acceleration of the virtual leader vehicle, , and respectively represent the time derivative of the position, speed and acceleration of the virtual leader vehicle, represents the input of the virtual leader vehicle.

[0082] In an embodiment of the present application, the communication topology of the networked vehicles is modeled as a directed graph, and in the communication topology of the networked vehicles , the node represents a vehicle, the edge represents the information transmission relationship between vehicles, represents the adjacency matrix of the communication topology , represents the connection weight of the vehicle transmitting information to the vehicle , if there is a connection between the vehicle and the vehicle , then =1, otherwise =0; wherein, , , represent the total number of vehicles;

[0083] In the communication topology , the communication connection state between vehicles is described by the Laplacian matrix ; wherein the element in the Laplacian matrix represents:

[0084] ;

[0085] In the formula, represents the connection weight of the vehicle transmitting information to the vehicle , ;

[0086] In the communication topology In, through a diagonal matrix Describes the communication connection status between the introduced virtual pilot vehicle and other vehicles, when the vehicle When the virtual navigator vehicle can directly receive information from the virtual navigator vehicle, the virtual navigator vehicle sends information to the vehicle. Connection weight for transmitting information ,otherwise .

[0087] In this embodiment of the invention, a novel distributed observer is designed to estimate the state of the lead vehicle. Utilizing the vehicle... The interaction between them is used to estimate the status of the virtual navigator vehicle; specifically, the vehicle... The distributed observer is represented as:

[0088] ;

[0089] In the formula, Indicates vehicle Estimation of the virtual navigator vehicle's position. Indicates vehicle Estimation of the virtual navigator vehicle's position. Indicates vehicle Estimation of the virtual navigator vehicle's position. Indicates the location of the virtual navigator vehicle. Indicates vehicle To the vehicle Connection weight for transmitting information, Indicates virtual navigator vehicle to vehicle Connection weight for transmitting information, , and These represent the first, second, and third positive fixed convergence parameters, respectively. Represents the Jacobian matrix. This indicates the current maximum time for scoring points. This represents the dummy time variable in the integration process. , Indicates the total number of vehicles;

[0090] Indicates vehicle The derivative of the estimate of the virtual pilot vehicle's speed. Indicates vehicle Estimation of the speed of the virtual navigator vehicle, Indicates vehicle Estimation of the speed of the virtual navigator vehicle, Indicates the speed of the virtual navigator vehicle;

[0091] representing the vehicle derivative of the estimate of the virtual lead vehicle acceleration, representing the vehicle estimate of the virtual lead vehicle acceleration, representing the vehicle estimate of the virtual lead vehicle acceleration, representing the virtual lead vehicle acceleration.

[0092] In embodiments of the invention, a vehicle controller is represented as:

[0093] ;

[0094] ;

[0095] ;

[0096] wherein, representing the vehicle controller, representing the position error of the vehicle , representing the speed error of the vehicle , representing the acceleration error of the vehicle , is represented as a nonlinear variable constructed as a vehicle , is represented as a nonlinear variable constructed as a vehicle , , and represent first, second and third positive fixed parameters, respectively;

[0097] representing the vehicle position, representing the desired distance between the vehicle and the virtual lead vehicle, representing the speed of the vehicle , representing the acceleration of the vehicle , representing the difference between the acceleration of the vehicle and the virtual lead vehicle.

[0098] In embodiments of the invention, a double-integral noise-robust gradient neural dynamics model (DINRGND) is constructed, comprising:

[0099] transforming the vehicle platoon control problem into a nonlinear problem and constructing an error function corresponding thereto;

[0100] Based on the constructed error function, an auxiliary variable is constructed for enhancing the convergence of the system and suppressing the noise steady-state error;

[0101] The constructed auxiliary variable is derived, and the auxiliary variable and the derived auxiliary variable are substituted into the condition formula to obtain a constructed double-integral anti-noise gradient neural dynamics model.

[0102] In the embodiment, the above model construction process is specifically:

[0103] To solve the vehicle arrangement problem, the platoon control is converted into a nonlinear problem, and the expression is as follows:

[0104] ;

[0105] In the formula, denotes a nonlinear mapping, and is the solution of . First, the expression of the error function is:

[0106] ;

[0107] In the formula, denotes the communication transmission relationship matrix between vehicles. For the new GNN, the expression is:

[0108] ;

[0109] In the formula, the superscript denotes transposition. For the convenience of expression, let .

[0110] The auxiliary variable for enhancing the convergence of the system and suppressing the noise steady-state error is defined , and the expression is as follows:

[0111] ;

[0112] In the formula, and denote the convergence parameters

[0113] The derived auxiliary variable is denoted as:

[0114] ;

[0115] The condition formula is:

[0116] ;

[0117] In the formula, denotes the error function, , and These represent the first, second, and third positive fixed convergence parameters, respectively. Represents the Jacobian matrix function. This indicates the current maximum time for scoring points. This represents the dummy time variable in the integration process. This represents the auxiliary variable after differentiation.

[0118] Substituting the auxiliary variables and the differentiated auxiliary variables into the above conditional formula, we obtain the double integral noise-resistant gradient neurodynamic model as follows:

[0119] ;

[0120] In the formula, Represents the error function The time derivative.

[0121] In this embodiment of the invention, an application example of the above-described automatic control method for networked vehicle platooning is provided.

[0122] In this embodiment, the communication relationship between vehicles is defined as follows: Figure 2 As shown, this is represented as an undirected graph. The unweighted case is used here; if the vehicle... With vehicles Once a communication connection is established, record If the vehicle If information can be obtained directly from the lead vehicle, then it is set to... The corresponding experimental results are as follows: Figures 3-4 As shown, in Figures 3-4 In the experiment, the five following vehicles were able to synchronize their acceleration and speed with the virtual navigator vehicle after 5 seconds, and the distance between the vehicles remained at about 8 meters. Further analysis showed that the speed and acceleration of the following vehicle were always synchronized with the preceding vehicle, and their trajectories overlapped after a certain period of time, indicating that the communication link could effectively overcome the influence of noise.

[0123] In this embodiment, a directional communication topology is defined between vehicles, such as... Figure 5 As shown, the corresponding experimental results are as follows: Figures 6-7 As shown, in Figures 6-7 The speed and acceleration of the five following vehicles remained synchronized even in noisy conditions, and their synchronization distance remained consistently at the desired 8 meters. Figure 6 As shown, the vehicle's acceleration gradually stabilizes over time, while Figure 7 This displays the distance difference, speed, and acceleration difference between vehicles.

[0124] In the above examples, whether it is an undirected or directed communication topology, the vehicle can maintain a stable distance and synchronization state under noise interference, further proving the robustness and effectiveness of the proposed communication strategy.

[0125] In the above examples, the synchronization behavior of vehicles in a noisy environment under undirected and directed communication topologies is studied, and the experimental results show that although there is noise interference, the vehicles can achieve synchronization after a certain time and maintain a stable distance (the expected distance is 8m) with the virtual lead vehicle. Therefore, the method has strong robustness and can cope with the interference of the noisy environment, ensuring that the vehicles can still achieve effective coordinated control under complex conditions.

[0126] The principles and implementation manners of the present application are described by using specific examples in the present application, and the above examples are only used to help understand the method of the present application and its core idea; meanwhile, for those skilled in the art, according to the idea of the present application, the specific implementation manners and application ranges can be changed, and the above description should not be understood as a limitation of the present application.

[0127] Those skilled in the art will realize that the examples described herein are for the purpose of aiding the reader to understand the principles of the present application and should be understood as not limiting the scope of protection of the present application to such specific statements and examples. Those skilled in the art can make various other specific modifications and combinations according to the technical inspiration of the present application without departing from the essence of the present application, and these modifications and combinations are still within the scope of protection of the present application.

Claims

1. A networked vehicle platooning automatic control method based on noise-resistant gradient neurodynamics, characterized in that, include: Construct a longitudinal dynamics model of the vehicle; Construct a communication topology for connected vehicles and introduce a virtual navigator vehicle into the communication topology to provide a reference trajectory; For the longitudinal dynamics model of the vehicle, a distributed observer and controller are constructed. A double-integral noise-resistant gradient neurodynamic model was constructed and embedded into the controller; During vehicle platooning control, based on the constructed communication topology, the controller performs vehicle control under error attenuation and noise suppression. Through the integral compensation mechanism of the distributed observer, the state estimation results of the virtual navigator are dynamically corrected using historical error information, thereby realizing automatic control of networked vehicle platooning.

2. The automatic control method for networked vehicle platooning based on noise-resistant gradient neurodynamics according to claim 1, characterized in that, The vehicle longitudinal dynamics model is a dynamic model that incorporates external disturbances, and it is expressed as follows: ； In the formula, Indicates vehicle The time derivative of the position, Indicates vehicle The time derivative of the velocity, Indicates vehicle The time derivative of the acceleration, Indicates vehicle speed, Indicates vehicle acceleration, Represents the linearized vehicle The controller input, Indicates vehicle The noise interference received, among which, , , This represents the new input after feedback linearization. This represents disturbances and model uncertainties. This represents the inertial delay in the longitudinal dynamics of a vehicle.

3. The automatic control method for networked vehicle platooning based on noise-resistant gradient neurodynamics according to claim 1, characterized in that, The communication topology of the networked vehicle In the middle, node Indicates vehicle, side This indicates the information transmission relationship between vehicles. Represents the communication topology The adjacency matrix, Indicates vehicle To the vehicle Connection weight for transmitting information, if vehicle With vehicles If there is a connection, then =1, otherwise =0; where, , , Indicates the total number of vehicles; In the communication topology Through the Laplace matrix Describes the communication connection status between vehicles; where the elements in the Laplace matrix... Represented as: ； In the formula, Indicates vehicle To the vehicle Connection weight for transmitting information, ; In the communication topology In, through a diagonal matrix Describes the communication connection status between the introduced virtual pilot vehicle and other vehicles, when the vehicle When the virtual navigator vehicle can directly receive information from the virtual navigator vehicle, the virtual navigator vehicle sends information to the vehicle. Connection weight for transmitting information ,otherwise .

4. The automatic control method for networked vehicle platooning based on noise-resistant gradient neurodynamics according to claim 1, characterized in that, vehicle The distributed observer is represented as: ； In the formula, Indicates vehicle Estimation of the virtual navigator vehicle's position. Indicates vehicle Estimation of the virtual navigator vehicle's position. Indicates vehicle Estimation of the virtual navigator vehicle's position. Indicates the location of the virtual navigator vehicle. Indicates vehicle To the vehicle Connection weight for transmitting information, Indicates virtual navigator vehicle to vehicle Connection weight for transmitting information, , and These represent the first, second, and third positive fixed convergence parameters, respectively. Represents the Jacobian matrix. This indicates the current maximum time for scoring points. This represents the dummy time variable in the integration process. , Indicates the total number of vehicles; Indicates vehicle The derivative of the estimate of the virtual pilot vehicle's speed. Indicates vehicle Estimation of the speed of the virtual navigator vehicle, Indicates vehicle Estimation of the speed of the virtual navigator vehicle, Indicates the speed of the virtual navigator vehicle; Indicates vehicle The derivative of the estimate of the acceleration of the virtual navigator vehicle. Indicates vehicle Estimation of acceleration of the virtual navigator vehicle, Indicates vehicle Estimation of acceleration of the virtual navigator vehicle, This indicates the acceleration of the virtual navigator vehicle.

5. The automatic control method for networked vehicle platooning based on noise-resistant gradient neurodynamics according to claim 4, characterized in that, vehicle The controller is represented as: ； ； ； In the formula, Indicates vehicle The controller, Indicates vehicle Position error, Indicates vehicle Speed ​​error, Indicates vehicle acceleration error, Represented as a vehicle Constructed nonlinear variables, Represented as a vehicle Constructed nonlinear variables, , and These represent the first, second, and third positive fixed parameters, respectively. Indicates vehicle Location, Indicates vehicle The expected distance between the virtual pilot vehicle and the virtual pilot vehicle. Indicates vehicle speed, Indicates vehicle acceleration, Indicates vehicle The difference in acceleration between the virtual pilot vehicle and the virtual navigator vehicle.

6. The automatic control method for networked vehicle platooning based on noise-resistant gradient neurodynamics according to claim 1, characterized in that, Constructing a dual-integral noise-resistant gradient neurodynamic model, including: The vehicle formation control problem is transformed into a nonlinear problem, and its corresponding error function is constructed. Based on the constructed error function, auxiliary variables are constructed to enhance system convergence and suppress noise steady-state error; Differentiate the constructed auxiliary variables and substitute the auxiliary variables and the differentiated auxiliary variables into the condition formula to obtain the constructed double integral noise-resistant gradient neurodynamic model.

7. The automatic control method for networked vehicle platooning based on noise-resistant gradient neurodynamics according to claim 6, characterized in that, The auxiliary variables used to enhance system convergence and suppress noise steady-state error Represented as: ； The conditional formula is: ； In the formula, Represents the error function. , and These represent the first, second, and third positive fixed convergence parameters, respectively. Represents the Jacobian matrix function. This indicates the current maximum time for scoring points. This represents the dummy time variable in the integration process. This represents the auxiliary variable after differentiation.

8. The automatic control method for networked vehicle platooning based on noise-resistant gradient neurodynamics according to claim 7, characterized in that, The dual-integral noise-resistant gradient neurodynamic model is expressed as follows: ； In the formula, Represents the error function The time derivative.

Images