A knowledge transfer-based multi-vehicle longitudinal platoon robust control method and system

Abstract

The application belongs to the technical field of vehicle control, and provides a multi-vehicle longitudinal formation robust control method and system based on knowledge transfer, which comprises the following steps: constructing a generalized nonlinear multi-vehicle longitudinal system model of heterogeneous vehicles; mapping an original constrained tracking error to an unconstrained tracking error based on a dynamic design of a segmented performance function of the original constrained tracking error; constructing a coupled sliding mode surface based on the unconstrained tracking error, designing a multi-vehicle longitudinal formation robust controller and an uncertain parameter adaptive law; collecting input data and output data of a first following vehicle under the multi-vehicle longitudinal formation robust controller, and introducing historical brake measurements as additional input data; training a long short-term memory network and performing iterative optimization; and transferring the knowledge in the trained long short-term memory network to other following vehicle control systems. The application solves the problems of low transient response rate, high design complexity, and a large number of parameter configurations in the traditional method.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of vehicle control, and particularly relates to a multi-vehicle longitudinal platoon robust control method and system based on knowledge transfer. BACKGROUND

[0002] The statements in this section merely provide background information related to the present application and do not necessarily constitute prior art.

[0003] With the rapid development of intelligent transportation systems and autonomous driving technologies, multi-vehicle longitudinal platooning, as an efficient and energy-saving traffic organization form, has gradually become a research hotspot. The core goal of multi-vehicle longitudinal platooning is to enable multiple autonomous or semi-autonomous vehicles to maintain dynamic synchronization in a compact formation on the same lane through inter-vehicle cooperative control, significantly improving road utilization, reducing energy consumption, and reducing traffic accident risk. In existing intelligent transportation systems, multi-vehicle longitudinal platooning technology has been verified in multiple pilot projects, such as freight vehicle platooning and autonomous taxi platooning, which not only changes the vehicle operation mode but also triggers the innovation of traffic system management paradigm.

[0004] In implementing multi-vehicle longitudinal platooning, traditional control algorithms are model-based and distributed, while actual heterogeneous multi-vehicle longitudinal platooning has complex nonlinear dynamics and environmental disturbances, which leads to insufficient robustness during operation, making it inconvenient for algorithm configuration and application, specifically in the following aspects:

[0005] First, the change in acceleration will directly change the kinetic energy of the vehicle, and a small fluctuation in speed may cause a step change in acceleration through the nonlinear characteristics of the drive system, forming a complex dynamic feedback loop. This coupling relationship makes it difficult for traditional linearization control strategies to guarantee the global stability of the system, especially in platoon reorganization scenarios that require frequent acceleration / deceleration, where the timing coordination of control instructions and parameter tuning face significant difficulties.

[0006] Second, in actual road environments, due to complex environments such as road slopes, strong winds, and strong sand, vehicle dynamics are disturbed by external disturbances. Steep slopes can cause sudden changes in drive torque demand, leading to sudden drops or rises in vehicle acceleration; strong winds can cause lateral deviation of the vehicle, indirectly coupling longitudinal dynamics, producing additional resistance or thrust; strong sand and dust can reduce tire adhesion, causing a deviation between wheel speed and actual vehicle speed, exacerbating longitudinal speed control errors.

[0007] Third, traditional distributed multi-vehicle longitudinal platooning control algorithms require laborious control parameter configuration for each vehicle, and real-time parameter adjustment is required to respond to environmental disturbances. In scenarios where communication delays occur between vehicles or local network interruptions occur, the vulnerability of distributed parameter configuration is further amplified, significantly reducing system robustness.

[0008] Fourth, most of the existing collision avoidance techniques rely on qualitative safety indicators such as minimum safety distance, but lack quantitative evaluation of path tracking accuracy and real-time performance indicators. Although the mainstream convergence control method can achieve vehicle longitudinal platoon control response, its convergence time is highly dependent on manually set control parameters. SUMMARY

[0009] To overcome the above shortcomings of the prior art, the present application provides a multi-vehicle longitudinal platoon robust control method and system based on knowledge transfer, which uses a long short-term memory network machine learning algorithm to depict the nonlinear coupling relationship of vehicle dynamics, improves the robustness of multi-vehicle platoon operation, controls the vehicle spacing error within the performance range set by humans, and solves the problems of low transient response rate, high design complexity, and large number of parameter configurations in traditional multi-vehicle longitudinal platoon control methods.

[0010] To achieve the above purpose, one or more embodiments of the present application provide the following technical solutions:

[0011] The first aspect of the present application provides a multi-vehicle longitudinal platoon robust control method based on knowledge transfer, comprising:

[0012] A generalized nonlinear multi-vehicle longitudinal system model of heterogeneous vehicles is constructed, considering the nonlinear coupling dynamics of displacement, speed, acceleration and external disturbances;

[0013] Based on the original constrained tracking error dynamics, a segmented performance function is designed to map the original constrained tracking error to an unconstrained tracking error, achieving the preset transient and steady-state convergence performance;

[0014] Based on the unconstrained tracking error, a coupled sliding surface is constructed, a multi-vehicle longitudinal platoon robust controller and an uncertain parameter adaptive law are designed to achieve the basic control of multi-vehicle longitudinal platoon;

[0015] The input data and output data of the first following vehicle under the multi-vehicle longitudinal platoon robust controller are collected, and historical braking measurements are introduced as additional input data; the input data is normalized for pretreatment, and the output data is denormalized for processing, which is used to train the long short-term memory network and perform iterative optimization;

[0016] The knowledge in the trained long short-term memory network is transferred to the control system of other following vehicles to achieve multi-vehicle longitudinal platoon robust control based on long short-term memory network.

[0017] As a further technical solution, the generalized nonlinear multi-vehicle longitudinal system model is specifically:

[0018]

[0019] wherein, for time; respectively the position, velocity and derivative of the acceleration of the i-th vehicle, ; ; , respectively the velocity and acceleration of the i-th vehicle; ; denotes a coupled nonlinear dynamics of velocity and acceleration; is a control gain parameter; is a control input; denotes an external disturbance caused by the environment.

[0020] As a further technical solution, the piecewise performance function is:

[0021]

[0022] wherein, , is a design parameter, and , ; is a stabilization time of the original constrained tracking error .

[0023] As a further technical solution, the unconstrained tracking error is:

[0024]

[0025]

[0026] wherein, is the original constrained tracking error, the invertible function is

[0027]

[0028] wherein, is a scalar, .

[0029] As a further technical solution, transient and steady state convergence regions are delineated:

[0030]

[0031] By delineating transient and steady state convergence regions, a pre-specified transient and steady state convergence performance is achieved.

[0032] As a further technical solution, a coupled sliding surface constructed from the unconstrained tracking error is:

[0033]

[0034]

[0035] wherein, is the sliding surface of the first vehicle, is the sliding surface of the first vehicle, is the sliding surface of the first vehicle, is the sliding surface of the first vehicle, , , , and are design parameters and are positive numbers; is the derivative of the unconstrained tracking error .

[0036] As a further technical solution, the multi-vehicle longitudinal platoon robust controller and the uncertain parameter adaptive law is

[0037]

[0038]

[0039]

[0040] wherein, is the unknown uncertain parameter generated by the fuzzy system approximation of the nonlinear function; is the estimated value of the parameter ; is the derivative of the estimated value, that is, the obtained uncertain parameter adaptive law; , , and are all design parameters.

[0041] The second aspect of the present application provides a multi-vehicle longitudinal platoon robust control system based on knowledge transfer, comprising:

[0042] a model construction module configured to construct a generalized nonlinear multi-vehicle longitudinal system model of heterogeneous vehicles, comprehensively considering the nonlinear coupling dynamics of displacement, speed, acceleration and external disturbances;

[0043] an error conversion module configured to map the original constrained tracking error to an unconstrained tracking error based on the original constrained tracking error dynamic design of the segmented performance function, and realize the preset transient and steady-state convergence performance;

[0044] A sliding mode control module is configured to construct a coupled sliding mode surface based on the unconstrained tracking error, design a multi-vehicle longitudinal formation robust controller and an uncertain parameter adaptive law, and realize basic control of multi-vehicle longitudinal formation.

[0045] A network training module is configured to collect input data and output data of a first follower vehicle under the multi-vehicle longitudinal formation robust controller, introduce historical braking measurements as additional input data, perform normalization preprocessing on the input data and denormalization processing on the output data, train a long short-term memory network, and perform iterative optimization.

[0046] A knowledge transfer module is configured to transfer knowledge in the trained long short-term memory network to other follower vehicle control systems, and realize multi-vehicle longitudinal formation robust control based on the long short-term memory network.

[0047] The third aspect of the present application provides a computer-readable storage medium having a program stored thereon, the program being executed by a processor to implement the steps of the multi-vehicle longitudinal formation robust control method based on knowledge transfer according to the first aspect of the present application.

[0048] The fourth aspect of the present application provides an electronic device comprising a memory, a processor, and a program stored on the memory and executable on the processor, wherein the processor implements the steps of the multi-vehicle longitudinal formation robust control method based on knowledge transfer according to the first aspect of the present application when executing the program.

[0049] The above one or more technical solutions have the following beneficial effects:

[0050] The present application constructs a generalized nonlinear vehicle system model by considering displacement, speed, acceleration, nonlinear coupling dynamics and external disturbances of heterogeneous vehicles.

[0051] The present application proposes a coupled sliding mode surface, which fully utilizes the communication information of bidirectional vehicles and leader vehicles, and effectively suppresses the disturbance problem caused by changes in the external environment of vehicles.

[0052] The present application uses input and output data generated by the multi-vehicle longitudinal formation robust controller to train a long short-term memory network, and configures the trained long short-term memory network to the entire multi-vehicle longitudinal control system, without the need to design a controller for each vehicle's longitudinal formation system, thereby improving the transient response rate of the control scheme implementation, reducing the design complexity, effectively reducing the parameter settings of the multi-vehicle longitudinal control system, and improving the stability and tracking accuracy of the multi-vehicle longitudinal formation.

[0053] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0054] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0055] Figure 1 This is a flowchart of the method in the first embodiment.

[0056] Figure 2 This is a schematic diagram of the displacement curves of the leader vehicle and the following vehicles 1-3 in the first embodiment;

[0057] Figure 3 This is a schematic diagram of the speed curves of the lead vehicle and follower vehicles 1-3 in the first embodiment;

[0058] Figure 4 This is a schematic diagram of the acceleration curves of the lead vehicle and follower vehicles 1-3 in the first embodiment;

[0059] Figure 5 This is a schematic diagram of the vehicle spacing error curves for following vehicles 1-3 in the first embodiment.

[0060] Figure 6 for Figure 5 Enlarged schematic diagram of the vehicle spacing error curve 2-3s for vehicles 1-3 following each other. Detailed Implementation

[0061] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0062] It should be noted that the terminology used herein is for the purpose of describing particular implementations only and is not intended to limit the exemplary implementations of the present invention.

[0063] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0064] The rise of multi-vehicle longitudinal platoon is driven by multiple needs. First, platooning can reduce the overall energy consumption of the platoon by reducing air resistance for the leading vehicle and following the energy-saving mode for the trailing vehicle, especially in high-speed freight scenarios. Second, urban traffic congestion is becoming increasingly severe, and platooning can improve traffic efficiency by precisely controlling the vehicle distance under existing road conditions. In addition, the maturity of automatic driving technologies, such as perception, decision-making algorithms, and inter-vehicle communication, provides a technical foundation for platooning, enabling real-time sharing of speed, position, and other data between vehicles to achieve multi-vehicle coordination.

[0065] However, when using traditional model-based, distributed control algorithms, the coupling relationship between acceleration and speed makes it difficult for traditional linear control strategies to guarantee the global stability of the system. The actual road environment exacerbates the longitudinal speed control error, requiring laborious control parameter configuration for each vehicle and real-time parameter adjustment, and lacking quantitative evaluation of path tracking accuracy and real-time performance indicators.

[0066] Therefore, the present application proposes a multi-vehicle longitudinal platoon robust control method to improve transient response rate and reduce design complexity and the number of parameter configurations.

[0067] Embodiment one

[0068] As shown in Figure 1 The present embodiment discloses a multi-vehicle longitudinal platoon robust control method based on knowledge transfer; comprising:

[0069] S1: Construct a generalized nonlinear multi-vehicle longitudinal system model for heterogeneous vehicles, considering the nonlinear coupling dynamics of displacement, speed, and acceleration, as well as external disturbances;

[0070] S2: Design a segmented performance function based on the original constrained tracking error dynamics, map the original constrained tracking error to an unconstrained tracking error, and achieve the preset transient and steady-state convergence performance;

[0071] S3: Construct a coupled sliding surface based on the unconstrained tracking error, design a multi-vehicle longitudinal platoon robust controller and an uncertain parameter adaptive law, and realize the basic control of multi-vehicle longitudinal platoon;

[0072] S4: Collect the input data and output data of the first following vehicle under the multi-vehicle longitudinal platoon robust controller, introduce historical braking measurements as additional input data; normalize the input data for preprocessing, and denormalize the output data for processing, which are used to train the long short-term memory network and perform iterative optimization;

[0073] S5: Transfer the knowledge in the trained long short-term memory network to the control systems of other following vehicles, and realize multi-vehicle longitudinal platoon robust control based on the long short-term memory network.

[0074] In S1, considering the diversity of the number and characteristics of vehicles in the longitudinal driving platoon, and incomplete state information often encountered in real scenarios, the embodiment adopts a third-order generalized nonlinear multi-vehicle longitudinal system model containing coupled nonlinearities and external disturbances to describe the motion of a single vehicle in the longitudinal driving platoon, and the generalized nonlinear multi-vehicle longitudinal system model is specifically:

[0075]

[0076] wherein, is time; are the derivatives of the position, velocity and acceleration of the i-th vehicle, respectively; are the velocity and acceleration of the i-th vehicle, respectively; represents the coupled nonlinear dynamics of the velocity and acceleration; is a control gain parameter; is a control input; represents external disturbances caused by the environment, such as road slopes, strong winds, strong sand and other complex environments. In S2, the piecewise performance function based on the original constrained tracking error design is:

[0077]

[0078]

[0079] wherein, is a positive, decreasing performance function, are design parameters, and , ; is the stable time of the original constrained tracking error , and the initial value of the tracking error satisfies .

[0080] The piecewise performance function contains performance constraints on the original constrained tracking error in both transient and steady-state stages, and further contains a preset convergence time , and the tracking error can be constrained in transient and steady-state performance under the action of such a performance function, and converges to the steady-state region within the preset time .

[0081] The distance between the two adjacent vehicles and the original constrained tracking error are defined as:

[0082] ​​​​​​

[0083]

[0084]

[0085] where, is the displacement of the th vehicle, is the displacement of the th vehicle, is the length of the th vehicle, is the desired safety gap, , and are design parameters, is the desired displacement trajectory of the th vehicle, is the speed of the th vehicle, is the speed of the leader vehicle, is the displacement of the leader vehicle, is the length of the leader vehicle, is the length of the th follower vehicle.

[0086] The original constrained tracking error is equivalently converted to an unconstrained tracking error as:

[0087]

[0088]

[0089] where, is the original constrained tracking error, the invertible function is

[0090]

[0091] where, is a positive scalar.

[0092] The transient and steady-state convergence regions are artificially defined as:

[0093]

[0094] The preset transient and steady-state convergence performance is achieved by defining the transient and steady-state convergence regions.

[0095] For the generalized nonlinear multi-vehicle longitudinal system model, a multi-vehicle longitudinal formation robust control algorithm is constructed based on the unconstrained tracking error.

[0096] Specifically, in S3, to ensure the stability of the longitudinal platoon, a coupled sliding mode surface based on unconstrained tracking error is designed :

[0097]

[0098]

[0099] where, is the sliding mode surface of the i-th vehicle, is the sliding mode surface of the i-th vehicle, , , , , , and are design parameters and are positive; is the derivative of the unconstrained tracking error ;

[0100] The derivative of the coupled sliding mode surface is:

[0101]

[0102] Since the parameter is uncertain, it needs to be obtained through online estimation. The adaptive law of the uncertain parameter is:

[0103]

[0104]

[0105] where, is a nonlinear function that uses a fuzzy system to approximate the unknown uncertain parameter and is related to the norm of the fuzzy weight vector, , is the fuzzy weight vector; is the estimated value of the parameter , which is calculated online through the adaptive law; is the derivative of the estimated value, which is also the obtained adaptive law of the uncertain parameter; and are both positive design parameters.

[0106] Based on the coupled sliding mode surface and the estimated value of the parameter , the updated multi-vehicle longitudinal platoon robust controller is represented as:

[0107]

[0108] where, positive design parameters are represented.

[0109] In S4, the multi-vehicle longitudinal robust controller is used for longitudinal platoon control of the first follower vehicle, and input-output data for training the long short-term memory network are generated.

[0110] All learning parameters are obtained through the network training process, which involves iterative optimization of input-output matching with the mean square error cost function. Specifically, during the training process, the input data is transmitted through the gating structure of the long short-term memory network, and the hidden state and memory cell at the current time step are calculated. After the forward propagation generates the network output, the difference between the predicted result and the true value is quantified by the mean square error cost function. Through gradient descent method, the Adam optimizer adjusts the parameter value according to the learning rate, and finds the global minimum value by minimizing the mean square error cost function. The long short-term memory network effectively alleviates the problem of gradient vanishing or explosion through the design of sigmoid activation function and cell state of the gating mechanism.

[0111] The input data for training the long short-term memory network is the coupled sliding mode surface measurement , the historical braking measurement As additional input data, the value indicates the braking situation of the first follower vehicle, which is consistent with the principle of nonlinear autoregressive exogenous model. At the same time, the relevant output measurement of the long short-term memory network is the predicted braking .

[0112] Before training the long short-term memory network, the coupled sliding mode surface measurement and the historical braking measurement are normalized to meet the standard normal distribution. After processing by the long short-term memory network, the predicted braking is adjusted to its normal range, and the trained network is finally used to realize the longitudinal tracking control of the first follower vehicle and the leader vehicle.

[0113] In S5, in order to transfer the knowledge contained in the long short-term memory network-based multi-vehicle longitudinal platoon robust controller trained and used to control the first follower vehicle (the first vehicle following the leader) to the control system of the remaining vehicles in the multi-vehicle platoon, it includes:

[0114] Without retraining the network, the knowledge in the trained long short-term memory network can be transferred to the longitudinal vehicle control loop in the longitudinal platoon. Because the label data exists in the source domain (the first follower vehicle control loop) and the target domain (the remaining follower vehicle control loop), it can be considered as an inductive transfer learning problem, thereby realizing the generalization across domains.

[0115] In the remaining longitudinal vehicle control loops, the input of the long short-term memory network corresponds to the coupling sliding mode surface measurement , The corresponding long short-term memory network related output measurement is the predicted braking , In addition, considering that the robust control structure based on the long short-term memory network is evolved from the spacing control loop of the first follower vehicle, the historical braking measurement is taken as additional network input data, while following the principle of nonlinear autoregressive exogenous model.

[0116] In this embodiment, in order to prove the effectiveness of this embodiment, the following simulation test is carried out for verification:

[0117] Consider a group of heterogeneous vehicle platoon consisting of 3 follower vehicles and 1 leader vehicle, the initial values of the system states are set as shown in Table 1:

[0118] Table 1 Initial values of system states

[0119]

[0120] In this simulation experiment, the ideal vehicle spacing is set to 8 meters, and the external disturbance is modeled by the function , which is the motor inertia delay.

[0121] The effectiveness of the robust control method for multi-vehicle longitudinal platoon based on knowledge transfer proposed in this embodiment is verified by using MATLAB and Simulink simulation software, and Figure 2 , Figure 3 , Figure 4 and Figure 5 are obtained.

[0122] In the displacement tracking trajectory snapshots of the leader vehicle and follower vehicles 1-3 shown in Figure 2 , it can be observed that the trained robust control system based on long short-term memory network of the first follower vehicle can achieve safe spacing tracking of follower vehicles 2 and 3, with dynamic adaptation and generalization ability.

[0123] In the speed snapshots of the leader vehicle, follower vehicles 1-3 shown in Figure 3 , it can be observed that this method can maintain the consistency of the speed of all vehicles.

[0124] In the acceleration snapshots of the leader vehicle, follower vehicles 1-3 shown in Figure 4 , it can be observed that this method can maintain the consistency of the acceleration of all vehicles.

[0125] In Figure 5 The vehicle spacing tracking error trajectory of the follower vehicles 1-3 is shown (the performance boundary is indicated by a black dotted line in the figure), which shows that all the follower vehicles in the vehicle platoon can converge to the preset steady-state performance region within 3 seconds under the proposed robust control framework based on the long short-term memory network.

[0126] Figure 6 To Figure 5 The convergence amplification diagram of the vehicle spacing tracking error trajectory 2-3s of the follower vehicles 1-3 is shown, and the corresponding line color is referred to Figure 5 .

[0127] Embodiment two

[0128] The embodiment discloses a multi-vehicle longitudinal platoon robust control system based on knowledge transfer, comprising:

[0129] The model construction module is configured to: construct a generalized nonlinear multi-vehicle longitudinal system model of heterogeneous vehicles, comprehensively considering the nonlinear coupling dynamics of displacement, speed and acceleration and external disturbances;

[0130] The error conversion module is configured to: design a segmented performance function based on the original constrained tracking error dynamics, map the original constrained tracking error to an unconstrained tracking error, and realize the preset transient and steady-state convergence performance;

[0131] The sliding mode control module is configured to: construct a coupled sliding mode surface based on the unconstrained tracking error, design a multi-vehicle longitudinal platoon robust controller and an uncertain parameter adaptive law, and realize the basic control of the multi-vehicle longitudinal platoon;

[0132] The network training module is configured to: collect the input data and output data of the first follower vehicle under the multi-vehicle longitudinal platoon robust controller, and introduce historical braking measurements as additional input data; normalize the input data for pretreatment, and denormalize the output data for processing, which are used to train the long short-term memory network and perform iterative optimization;

[0133] The knowledge transfer module is configured to: transfer the knowledge in the trained long short-term memory network to the control system of other follower vehicles, and realize the multi-vehicle longitudinal platoon robust control based on the long short-term memory network.

[0134] Embodiment three

[0135] The purpose of the embodiment is to provide a computer readable storage medium.

[0136] The computer readable storage medium has a computer program stored thereon, and the program is executed by a processor to realize the steps in the multi-vehicle longitudinal platoon robust control method based on knowledge transfer described in embodiment one of the disclosure.

[0137] Embodiment Four

[0138] An electronic device is provided.

[0139] The electronic device includes a memory, a processor, and a program stored in the memory and executable on the processor, and the processor implements the steps in the method for robust control of multi-vehicle longitudinal platoon based on knowledge transfer according to the embodiment of the present disclosure.

[0140] The steps involved in the devices of embodiments two, three, and four above correspond to the method of embodiment one, and the specific implementation can be seen in the relevant description of embodiment one. The term "computer-readable storage medium" should be understood to include a single medium or multiple media that store one or more sets of instructions; it should also be understood to include any medium that is capable of storing, encoding, or carrying the set of instructions for execution by a processor and that causes the processor to perform any one of the methods of the present disclosure.

[0141] Those skilled in the art should understand that the modules or steps of the present disclosure described above can be implemented by a general computer device, and alternatively, they can be implemented by program codes executable by a computing device, so that they can be stored in a storage device and executed by a computing device, or they can be made into individual integrated circuit modules, or a plurality of modules or steps among them can be made into a single integrated circuit module. The present disclosure is not limited to any specific combination of hardware and software.

[0142] The specific embodiments of the present disclosure are described above in conjunction with the accompanying drawings, but are not a limitation on the scope of protection of the present disclosure, and those skilled in the art should understand that various modifications or changes made by those skilled in the art on the basis of the technical solutions of the present disclosure without creative labor are still within the scope of protection of the present disclosure.

Claims

1. A knowledge transfer-based multi-vehicle longitudinal platoon robust control method, characterized in that, The method comprises the following steps: A generalized nonlinear multi-vehicle longitudinal system model of heterogeneous vehicles is constructed, considering the nonlinear coupling dynamics of displacement, speed and acceleration and external disturbances; A piecewise performance function is designed based on the original constrained tracking error dynamics, the original constrained tracking error is mapped into an unconstrained tracking error, and preset transient and steady-state convergence performance is achieved; A coupling sliding mode surface is constructed based on the unconstrained tracking error, a multi-vehicle longitudinal formation robust controller and an uncertain parameter adaptive law are designed, and basic control of multi-vehicle longitudinal formation is achieved; Input data and output data of the first follower vehicle under the multi-vehicle longitudinal formation robust controller are collected, and historical braking measurements are introduced as additional input data; The input data is normalized and preprocessed, and the output data is denormalized for training a long short-term memory network and iterative optimization; The knowledge in the trained long short-term memory network is transferred to other follower vehicle control systems to achieve multi-vehicle longitudinal formation robust control based on the long short-term memory network.

2. The knowledge transfer based robust control method for multi-vehicle longitudinal platoon according to claim 1, wherein, The generalized nonlinear multi-vehicle longitudinal system model is specifically: where t is time; are the position, velocity and acceleration of the i-th vehicle, respectively, are the position, velocity and acceleration of the i-th vehicle, respectively, ; , are the position, velocity and acceleration of the i-th vehicle, respectively, are the position, velocity and acceleration of the i-th vehicle, respectively, denotes a coupled nonlinear dynamics of velocity and acceleration; is a control gain parameter; is a control input; denotes external disturbances caused by the environment.

3. The knowledge transfer based robust control method for multi-vehicle longitudinal platoon according to claim 1, wherein, The piecewise performance function is: wherein , is a design parameter, and , ; is the settling time of the original constrained tracking error .

4. The knowledge transfer based multi-vehicle longitudinal platoon robust control method of claim 3, wherein, The unconstrained tracking error Is: wherein the original constrained tracking error, the invertible function is wherein is a scalar, .

5. The knowledge transfer based robust control method for multi-vehicle longitudinal platoon according to claim 4, wherein, The transient and steady-state convergence regions are divided: The preset transient and steady-state convergence performance is achieved by dividing the transient and steady-state convergence regions.

6. The knowledge transfer based robust control method for multi-vehicle longitudinal platoon according to claim 4, wherein, Coupled sliding mode surface constructed from unconstrained tracking error is: wherein is the slip surface of the th vehicle, is the slip surface of the th vehicle, , , , and are design parameters and are positive; is the derivative of the unconstrained tracking error .

7. The knowledge transfer based robust control method for multi-vehicle longitudinal platoon according to claim 6, wherein, The multi-vehicle longitudinal platoon robust controller and an uncertain parameter adaptive law is where unknown uncertain parameters resulting from the approximation of the nonlinear function by the fuzzy system; an estimate of the parameter ; is the derivative of the estimate, i.e. the obtained adaptive law for the uncertain parameter , , and are design parameters.

8. A knowledge transfer based robust control system for multi-vehicle longitudinal platoon, characterized in that, The method comprises the following steps: A model construction module is configured to construct a generalized nonlinear multi-vehicle longitudinal system model of heterogeneous vehicles, considering the nonlinear coupling dynamics of displacement, speed and acceleration and external disturbances; An error conversion module is configured to design a piecewise performance function based on the original constrained tracking error dynamics, map the original constrained tracking error into an unconstrained tracking error, and achieve preset transient and steady-state convergence performance; A sliding mode control module is configured to construct a coupling sliding mode surface based on the unconstrained tracking error, design a multi-vehicle longitudinal formation robust controller and an uncertain parameter adaptive law, and achieve basic control of multi-vehicle longitudinal formation; A network training module is configured to collect input data and output data of the first follower vehicle under the multi-vehicle longitudinal formation robust controller, introduce historical braking measurements as additional input data, normalize and preprocess the input data, denormalize the output data for training a long short-term memory network, and perform iterative optimization; A knowledge transfer module is configured to transfer the knowledge in the trained long short-term memory network to other follower vehicle control systems to achieve multi-vehicle longitudinal formation robust control based on the long short-term memory network.

9. A computer-readable storage medium having stored thereon a program, characterized in that, The program is executed by the processor to implement the steps of the multi-vehicle longitudinal formation robust control method based on knowledge transfer according to any one of claims 1-7.

10. An electronic device comprising a memory, a processor, and a program stored on the memory and executable on the processor, characterized in that, The processor executes the program to implement the steps of the multi-vehicle longitudinal formation robust control method based on knowledge transfer according to any one of claims 1-7.

Images