Self-adaptive cruise control method and device based on online self-adaptive learning and medium

Abstract

The invention discloses a self-adaptive cruise control method and device based on online self-adaptive learning and a medium, and belongs to the technical field of self-adaptive control. The method comprises the following steps: constructing a mathematical model of the adaptive cruise control system, and defining an H infinite control problem related to the adaptive cruise control system; defining the state of the adaptive cruise control system based on a preset performance function algorithm, and constructing an online strategy learning algorithm; and solving an H infinity control problem based on an online strategy learning algorithm to determine an optimal control strategy. According to the method, the technical effect of improving the adaptability and robustness of the ACC system in a complex dynamic environment is achieved.

Description

Technical Field

[0001] The present application relates to the technical field of adaptive control, and in particular to an adaptive cruise control method, device and medium based on online adaptive learning. Background Art

[0002] The Adaptive Cruise Control (ACC) system is an advanced driver assistance system that adds a function of maintaining a safe distance from the vehicle in front based on the traditional cruise control. The system can monitor the acceleration, speed, distance and relative speed of the host vehicle (i.e. the vehicle equipped with the ACC system) in real time, and accurately control the throttle and brake system of the host vehicle through the built-in controller. The ACC system consists of two core controllers: the upstream controller and the downstream controller. The upstream controller is responsible for calculating the ideal host vehicle acceleration signal based on the collected real-time data, while the downstream controller is responsible for converting this ideal acceleration signal into actual vehicle acceleration, which is achieved by accurately adjusting the throttle and brake.

[0003] Current adaptive cruise control systems have limitations in dealing with dynamic uncertainties (such as changes in road conditions, difficulty in predicting other vehicles’ behavior, etc.) and external disturbances (such as sudden deceleration of drivers and passengers). These uncertainties and disturbances may cause the system to be unable to adjust the vehicle status in a timely and accurate manner, thus affecting driving safety and ride comfort.

[0004] Therefore, how to improve the adaptability and robustness of the ACC system in complex dynamic environments has become a technical problem that needs to be solved urgently. Summary of the invention

[0005] The embodiments of the present application provide an adaptive cruise control method, device and medium based on online adaptive learning, which are used to solve the following technical problem: how to improve the adaptability and robustness of the ACC system in a complex dynamic environment.

[0006] In a first aspect, an embodiment of the present application provides an adaptive cruise control method based on online adaptive learning, the method comprising: constructing a mathematical model of an adaptive cruise control system, and defining an H-infinity control problem related to the adaptive cruise control system; defining the state of the adaptive cruise control system based on a preset performance function algorithm, and constructing an online strategy learning algorithm; solving the H-infinity control problem based on the online strategy learning algorithm to determine the optimal control strategy.

[0007] In one implementation of the present application, a mathematical model of an adaptive cruise control system is established, and an H-infinity control problem related to the adaptive cruise control system is defined, specifically including: The formula of the mathematical model of the adaptive cruise control system is: (1) Among them, matrix A is the state matrix, matrix B is the input matrix, and matrix D is the interference matrix. is the state vector, is the control input, For external interference, Indicates the headway time difference, that is, the time interval between the main vehicle and the target vehicle. represents the time constant related to the dynamic response of the vehicle, , , Indicates the ideal stopping distance of the host vehicle. Indicates the actual stopping distance of the host vehicle. , Indicates the speed of the host vehicle. is the speed of the target vehicle, , is the main vehicle acceleration, , is the input desired acceleration, w is the disturbance of the adaptive cruise control system; The formula for the H-infinity control problem of the adaptive cruise control system is: (2) in, is the positive definite solution of the formula, * represents the number of iterations, Q and R are both positive definite weight matrices, is a small positive real number, It is used to limit external interference to the output of the adaptive cruise control system within a controllable range.

[0008] In one implementation of the present application, the state of the adaptive cruise control system is defined based on a preset performance function algorithm, and an online strategy learning algorithm is constructed, specifically including: Build offline policy learning algorithms; constructing a planning performance function to constrain the offline policy learning algorithm, and processing the offline policy learning algorithm based on the planning performance function to determine an initial state of the online policy learning algorithm; Construct an online policy learning algorithm based on the initial state.

[0009] In one implementation of the present application, an offline strategy learning algorithm is constructed, specifically including: The offline policy learning algorithm is expressed by the following formula: Set up an initial stable matrix ; based on , according to formula (3) : (3) in, To describe the matrix of the dynamic changes of the adaptive cruise control system, To calculate the control gain matrix of the control input according to the state of the adaptive cruise control system, is the disturbance gain matrix that calculates the disturbance input according to the state of the adaptive cruise control system, and are the state and control weight matrices respectively; According to formula (4), the initial optimal solution is obtained : set up , , ; Take the limit: , , ; (4) in, is the identity matrix, Indicates The control gain matrix of the iteration is, Indicates the adjustment matrix during iteration Parameters, Indicates The positive solution of the iteration is ; for Other situations: (5) in, For the matrix The characteristic value of For the matrix The real part of the eigenvalue, A matrix for adjusting and greater than real number; For each matrix satisfy:

[0010] Thereby constructing an offline policy learning algorithm.

[0011] In one implementation of the present application, a planning performance function is constructed to constrain the offline policy learning algorithm, and the offline policy learning algorithm is processed based on the planning performance function to determine the initial state of the online policy learning algorithm, specifically including: Construct a smooth function with positive decreasing properties , smooth function The formula is: (7) in , ,satisfy conditions, and As the lower and upper bounds respectively, limit the state of the adaptive cruise control system , is a positive constant greater than 0; based on and Limiting Adaptive Cruise Control System Status , to obtain the transformed adaptive cruise control system state .

[0012] In one implementation of the present application, an online strategy learning algorithm is constructed, specifically including: Based on formula (2) and the transformed adaptive cruise control system state ,definition ,definition , define the cost function as ; Two exploration signals used to collect data and Incorporating them into the control input and external disturbance respectively, formula (1) can be rewritten as: (8) Derivative the cost function: (9) in is the cost function of the adaptive cruise control system, which represents the target performance to be achieved by the adaptive cruise control system. Indicates that the adaptive cruise control system is in state , control input and interference input The instantaneous cost of the joint action; To collect adaptive cruise control system data, the following matrix is ​​defined:

[0013]

[0014]

[0015] (10) in, ; Define column full rank According to formula (11), we get , and ; (11) in, Indicates that it contains Equation Solution The estimated value vector, and They represent the control gain matrices and the interference gain matrix The vectorized form of definition is a positive constant greater than 0, and ,but: (12) in, Adaptive cruise control system status , control input and interference input The dimension of Based on the above steps, a preliminary online policy learning algorithm is constructed; among them, the preliminary online policy learning algorithm needs to determine the initial matrix ; Remove the initial matrix , to rewrite formula (1) to determine the online policy learning algorithm.

[0016] In one implementation of the present application, formula (1) is rewritten to determine the online strategy learning algorithm, specifically including: Rewrite formula (1) to generate formula (13): (13) in, , The state matrix of the adaptive cruise control system is adjusted to describe the dynamic changes of the system state; definition and , derive and rewrite formula (9) to obtain formula (14): (14) Through the matrix Solving and Include , and Information problems, using replace , then select , to obtain formula (15): (15) in, , To adjust the weight matrix A positive number, and Respectively represent the maximum eigenvalue and minimum eigenvalue of the matrix; Based on the above steps, the adaptive cruise control data is organized using metrics and the processed data is imported into the matrix tool; , , and After normalization, we get formula (16):

[0017]

[0018]

[0019]

[0020] (16) By column full rank matrix and , solve : (17) Define a column-full rank matrix: , , , to calculate , and : (18) Based on the above steps, an online policy learning algorithm is obtained.

[0021] In one implementation of the present application, solving the H-infinity control problem based on an online strategy learning algorithm to determine the optimal control strategy specifically includes: Initialize the online policy learning algorithm to , , and ; Repeat control input and interference Collecting data from adaptive cruise control systems ; when satisfy , then according to formula (17) ; According to formula (18), , and ; choose Satisfy the conditions of formula (15) and let Until and ; set up , , and ; According to formula (3), we can solve ; set up , , collect adaptive cruise control system information until formula (12) is satisfied; According to formula (11), we can solve , and ,until and Return , , , to determine the optimal control strategy.

[0022] In a second aspect, an embodiment of the present application also provides an adaptive cruise control device based on online adaptive learning, characterized in that the device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can: construct a mathematical model of the adaptive cruise control system, and define an H-infinity control problem related to the adaptive cruise control system; define the state of the adaptive cruise control system based on a preset performance function algorithm, and construct an online policy learning algorithm; solve the H-infinity control problem based on the online policy learning algorithm to determine the optimal control strategy.

[0023] In a third aspect, an embodiment of the present application further provides a non-volatile computer storage medium for adaptive cruise control based on online adaptive learning, storing computer executable instructions, characterized in that the computer executable instructions are configured to: construct a mathematical model of an adaptive cruise control system, and define an H-infinity control problem related to the adaptive cruise control system; define the state of the adaptive cruise control system based on a preset performance function algorithm, and construct an online policy learning algorithm; solve the H-infinity control problem based on the online policy learning algorithm to determine the optimal control strategy.

[0024] The embodiments of the present application provide an adaptive cruise control method, device and medium based on online adaptive learning, which at least include the following technical effects: The mathematical model of the adaptive cruise control system was constructed, and the related H-infinity control problem was clearly defined, laying a solid theoretical foundation for the performance optimization of the system. An online strategy learning algorithm with real-time learning and optimization capabilities was constructed. By continuously analyzing real-time data, the algorithm can dynamically adjust the control strategy to adapt to the ever-changing traffic environment. This adaptability and flexibility enables the system to track the vehicle ahead more stably and accurately, effectively avoiding safety hazards caused by sudden changes in traffic conditions. By applying control theory and optimization algorithms, the optimal performance of the adaptive cruise system can be determined to ensure that the vehicle maintains a safe distance and achieves high efficiency and energy saving during cruising. It not only improves the driving comfort and convenience, but also improves the robustness and stability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings: Figure 1 A flow chart of an adaptive cruise control method based on online adaptive learning provided in an embodiment of the present application; Figure 2 This is the structural diagram of the adaptive cruise control system for this application; Figure 3 This is a convergence effect diagram of the adaptive cruise control system when PPF exists in this application; Figure 4 This is a convergence effect diagram of the adaptive cruise control system when there is no PPF in this application; Figure 5 The maximum value change rate in this application and The strategy calculation value of ; Figure 6 In this application, the online strategy learning algorithm strategy error; Figure 7 In this application, the online strategy learning algorithm strategy error; Figure 8 In this application, the online strategy learning algorithm strategy error; Fig. 9 This is a rendering of an embodiment of the present application; Fig. 9 (a) is a comparison diagram of the actual distance and the ideal distance of the embodiment of the present application; Fig. 9 (b) is a comparison diagram of the main vehicle speed and the target vehicle speed in the embodiment of the present application; Fig.10 It is a rendering of the prior art; Fig.10 (a) is a comparison diagram of the actual distance and the ideal distance of the prior art; Fig.10 (b) is a comparison diagram of the speed of the main vehicle and the speed of the target vehicle in the prior art; Fig.11 A schematic diagram of trajectory errors under PPF limitation in the embodiment of the present application and the prior art; Fig.11 (a) is a schematic diagram of the trajectory error under PPF limitation in an embodiment of the present application; Fig.11 (b) is a schematic diagram of the trajectory error of the prior art under PPF constraints; Fig.12 A comparison diagram of the integrated absolute error of an adaptive cruise control method based on online adaptive learning provided by an embodiment of the present application and the error of the prior art; Fig.13 A schematic diagram of the internal structure of an adaptive cruise control device based on online adaptive learning provided in an embodiment of the present application. DETAILED DESCRIPTION

[0026] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described below in combination with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0027] The embodiments of the present application provide an adaptive cruise control method, device and medium based on online adaptive learning, which are used to solve the following technical problem: how to improve the adaptability and robustness of the ACC system in a complex dynamic environment.

[0028] The technical solution proposed in the embodiments of the present application is described in detail below with reference to the accompanying drawings.

[0029] Figure 1 The present invention provides an adaptive cruise control flow chart based on online adaptive learning. Figure 1 As shown, an adaptive cruise control method based on online adaptive learning provided by an embodiment of the present application specifically includes the following steps: Step 1: Construct a mathematical model of the adaptive cruise control system and define the H-infinity control problem of the adaptive cruise control system.

[0030] Figure 2 This is a structural diagram of the adaptive cruise control system.

[0031] First, a mathematical model of the adaptive cruise control system needs to be constructed.

[0032] The formula of the mathematical model of the adaptive cruise control system is: (1) Among them, matrix A is the state matrix, matrix B is the input matrix, and matrix D is the interference matrix. is the state vector, is the control input, For external interference, Indicates the headway time difference, that is, the time interval between the main vehicle and the target vehicle. represents the time constant related to the dynamic response of the vehicle, , , Indicates the ideal stopping distance of the host vehicle. Indicates the actual stopping distance of the host vehicle. , Indicates the speed of the host vehicle. is the speed of the target vehicle, , is the main vehicle acceleration, , is the input desired acceleration, and w is the disturbance of the adaptive cruise control system.

[0033] Furthermore, an H-infinity control problem related to the mathematical model of the adaptive cruise control system is constructed.

[0034] The formula for the H-infinity control problem is: (2) in, is the positive definite solution of this formula, Q and R are both positive definite weight matrices, is a small positive real number, It is used to limit external interference to the output of the adaptive cruise control system within a controllable range.

[0035] Step 2: Define the state of the adaptive cruise control system based on a preset performance function algorithm and build an online strategy learning algorithm.

[0036] First, build an offline policy learning algorithm.

[0037] The offline policy learning algorithm is expressed by the following formula: Set up an initial stable matrix ; based on , according to formula (3) : (3) in To describe the matrix of the dynamic changes of the adaptive cruise control system, To calculate the control gain matrix of the control input according to the state of the adaptive cruise control system, is the disturbance gain matrix that calculates the disturbance input according to the state of the adaptive cruise control system, and are the state and control weight matrices respectively; According to formula (4), the initial optimal solution is obtained : set up , , ; Take the limit: , , ; (4) in, is the identity matrix, Indicates The control gain matrix of the iteration is, Indicates the adjustment matrix during iteration Parameters, Indicates The positive solution of the iteration is ; for Other situations: (5) in For the matrix The characteristic value of For the matrix The real part of the eigenvalue, A matrix for adjusting and greater than real number; For each matrix satisfy:

[0038] Thereby constructing an offline policy learning algorithm.

[0039] Further, a planning performance function is constructed to constrain the offline policy learning algorithm, and the offline policy learning algorithm is processed based on the planning performance function to determine an initial state of the online policy learning algorithm; Construct a smooth function with positive decreasing properties , smooth function The formula is: (7) in , ,satisfy conditions, and As the lower and upper bounds respectively, limit the state of the adaptive cruise control system , is a positive constant greater than 0; based on and Limiting Adaptive Cruise Control System Status , to obtain the transformed adaptive cruise control system state .

[0040] In a specific example, setting the parameters , , , when there is and When limiting, the state of the adaptive cruise control system The limiting effect of Figure 3 shown.

[0041] When there is no and Limiting Adaptive Cruise Control System Status When the adaptive cruise control system is in 1. 2. The tracking error of 3 is Figure 4 shown.

[0042] Furthermore, an online strategy learning algorithm is constructed based on the initial state.

[0043] Based on formula (2) and the transformed adaptive cruise control system state ,definition ,definition , define the cost function as ; Two exploration signals used to collect data and Incorporating them into the control input and external disturbance respectively, formula (1) can be rewritten as: (8) Derivative the cost function: (9) in is the cost function of the adaptive cruise control system, which represents the target performance to be achieved by the adaptive cruise control system. Indicates that the adaptive cruise control system is in state , control input and interference input The instantaneous cost of the joint action; To collect adaptive cruise control system data, the following matrix is ​​defined:

[0044]

[0045]

[0046] (10) in, ; Define column full rank According to formula (11), we get , and ; (11) in, Indicates that it contains Equation Solution The estimated value vector, and They represent the control gain matrices and the interference gain matrix The vectorized form of definition is a positive constant greater than 0, and ,but: (12) in, Adaptive cruise control system status , control input and interference input The dimension of Based on the above steps, a preliminary online policy learning algorithm is constructed; among them, the preliminary online policy learning algorithm needs to determine the initial matrix ; Remove the initial matrix , to rewrite formula (1) to generate formula (13): (13) in, , The state matrix of the adaptive cruise control system is adjusted to describe the dynamic changes of the system state; definition and , derive and rewrite formula (9) to obtain formula (14): (14) Through the matrix Solving and Include , and Information problems, using replace , then select , to obtain formula (15): (15) in, , To adjust the weight matrix A positive number, and Respectively represent the maximum eigenvalue and minimum eigenvalue of the matrix; Based on the above steps, the adaptive cruise control data is organized using metrics and the processed data is imported into the matrix tool; , , and After normalization, we get formula (16):

[0047]

[0048]

[0049]

[0050] (16) By column full rank matrix and , solve : (17) Define a column-full rank matrix: , , , to calculate , and : (18) Based on the above steps, an online policy learning algorithm is obtained.

[0051] Step 3: Solve the H-infinity control problem based on the online policy learning algorithm to determine the optimal performance of the adaptive cruise control system.

[0052] First, initialize the online policy learning algorithm to , , and ; Furthermore, repeated control input and interference Collecting data from adaptive cruise control systems ; Furthermore, when satisfy , then according to formula (17) ; Furthermore, according to formula (18), , and ; Further, select Satisfy the conditions of formula (15) and let Until and ; Furthermore, , , and ; Furthermore, according to formula (3), ; Furthermore, setting , , collect adaptive cruise control system information until formula (12) is satisfied; Furthermore, according to formula (11), , and ,until and Return , , , to determine the optimal control strategy.

[0053] In a specific example, the experimental vehicle is equipped with sensors such as millimeter-wave radar, lidar, and cameras to collect perception information of the surrounding environment. The on-board computing unit is equipped with a multimodal environment perception fusion algorithm to accurately measure the vehicle's position, speed, and acceleration information. The time constant of the lower controller of the adaptive cruise control system is set to , workshop time distance . Set a non-zero initial system state The goal of the control problem in adaptive cruise control is to transform the state vector Drive to zero. Choose the identity matrix as and ,set up , the solution of formula (3) is:

[0054] P* is the solution of the offline policy algorithm. The main function of P* is to determine whether the result of the online policy learning algorithm is consistent with the solution of the offline policy learning algorithm.

[0055] The initial matrix is ​​obtained through the online policy learning algorithm ,set up , , and introduce exploration signals into online policy learning algorithms and , in each iteration Increase by 0.1 The calculation result of is greater than 0. Once the conditions are met, continue to use the online strategy learning algorithm to solve . Figure 5 The maximum rate of change and .

[0056] choose As a steady-state feedback matrix for online policy learning algorithms . Figures 6 to 8 The calculation results in the algorithm show , and The convergence of .

[0057] The learning results of the tenth iteration for:

[0058] Based on the above iterative results, the approximate error is obtained The solution obtained from the online policy learning algorithm is close to the solution derived from Equation (3), which shows that it is feasible to use the online policy learning based method to solve the model-free H∞ control problem in the adaptive cruise control system.

[0059] Based on the learning results of the proposed online strategy learning algorithm, the tracking trajectory of the main vehicle is as follows: Fig. 9 and Fig.10 As shown, the trajectory diagrams of the distance and speed between the main vehicle and the target vehicle respectively demonstrate the superiority of the tracking effect of the method designed in the present application applied to the ACC system compared with the tracking effect of the reinforcement learning method used in the prior invention applied to the ACC system.

[0060] To further demonstrate the superiority of the proposed control method in terms of accuracy, it is compared with the reinforcement learning algorithm in the prior art. The distance error x1, speed error x2 and acceleration x3 of the host vehicle are as follows: Fig.11 and Fig.12 As shown: From the above results, it can be seen that the algorithm designed in the embodiment of the present application has excellent tracking accuracy. The model-free control method designed and adopted can learn from real-time input and output data, providing higher robustness and adaptability, especially for systems with uncertainty and dynamic changes.

[0061] The above is an embodiment of the method proposed in this application. Based on the same inventive concept, the embodiment of this application also provides an adaptive cruise control device based on online adaptive learning, whose structure is as follows: Fig.13 shown.

[0062] Fig.13 The internal structure diagram of an adaptive cruise control device based on online adaptive learning provided by the embodiment of the present application is as follows. Fig.13 As shown, the device includes: at least one processor 201; and, a memory 202 communicatively connected to the at least one processor; The memory 202 stores instructions that can be executed by at least one processor, and the instructions are executed by at least one processor 201 to enable at least one processor 201 to: Construct a mathematical model of an adaptive cruise control system and define the H-infinity control problem related to the adaptive cruise control system; construct a mathematical model of an adaptive cruise control system and define the H-infinity control problem related to the adaptive cruise control system; define the state of the adaptive cruise control system based on a preset performance function algorithm and construct an online policy learning algorithm; solve the H-infinity control problem based on the online policy learning algorithm to determine the optimal control strategy.

[0063] Some embodiments of the present application provide corresponding Figure 1 A non-volatile computer storage medium for an adaptive cruise control system based on online adaptive learning, storing computer executable instructions, wherein the computer executable instructions are set to: Construct a mathematical model of an adaptive cruise control system and define the H-infinity control problem related to the adaptive cruise control system; define the state of the adaptive cruise control system based on a preset performance function algorithm and construct an online policy learning algorithm; solve the H-infinity control problem based on the online policy learning algorithm to determine the optimal control strategy.

[0064] Each embodiment in this application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the IoT device and medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments.

[0065] The system and medium provided in the embodiments of the present application correspond one-to-one to the method. Therefore, the system and medium also have similar beneficial technical effects to the corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the system and medium will not be repeated here.

[0066] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0067] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0068] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0069] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0070] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0071] Memory may include non-permanent storage in a computer-readable medium, in the form of random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0072] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0073] It should also be noted that the terms "include", "comprising" and any other variations are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprising a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0074] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. An adaptive cruise control method based on online adaptive learning, characterized in that: The method comprises: Construct the mathematical model of adaptive cruise control system and define the H-infinity control problem of adaptive cruise control system; Defining the state of the adaptive cruise control system based on a preset performance function algorithm and constructing an online strategy learning algorithm; The H-infinity control problem is solved based on the online strategy learning algorithm to determine the optimal control strategy.

2. The adaptive cruise control method based on online adaptive learning according to claim 1, characterized in that: The mathematical model of adaptive cruise control system is established, and the H-infinity control problem of adaptive cruise control system is defined, including: The formula of the mathematical model of the adaptive cruise control system is: （1） Among them, matrix A is the state matrix, matrix B is the input matrix, and matrix D is the interference matrix. is the state vector, is the control input, For external interference, Indicates the headway time difference, that is, the time interval between the main vehicle and the target vehicle. represents the time constant related to the dynamic response of the vehicle, , , Indicates the ideal stopping distance of the host vehicle. Indicates the actual stopping distance of the host vehicle. , Indicates the speed of the host vehicle. is the speed of the target vehicle, , is the main vehicle acceleration, , is the input desired acceleration, w is the disturbance of the adaptive cruise control system; The formula of the H-infinity control problem of the adaptive cruise control system is: （2） in, is the positive definite solution of the formula, * represents the number of iterations, Q and R are both positive definite weight matrices, is a small positive real number, It is used to limit external interference to the output of the adaptive cruise control system within a controllable range.

3. The adaptive cruise control method based on online adaptive learning according to claim 1, characterized in that: Defining the state of the adaptive cruise control system based on a preset performance function algorithm and constructing an online strategy learning algorithm specifically includes: Build offline policy learning algorithms; Constructing a planning performance function to constrain the offline policy learning algorithm, and processing the offline policy learning algorithm based on the planning performance function to determine an initial state of the online policy learning algorithm; The online policy learning algorithm is constructed based on the initial state.

4. The adaptive cruise control method based on online adaptive learning according to claim 3, characterized in that: Build an offline policy learning algorithm, including: The offline strategy learning algorithm is expressed by the following formula: Set up an initial stable matrix ; based on , according to formula (3) : （3） in, To describe the matrix of the dynamic changes of the adaptive cruise control system, To calculate the control gain matrix of the control input according to the state of the adaptive cruise control system, is the disturbance gain matrix that calculates the disturbance input according to the state of the adaptive cruise control system, and are the state and control weight matrices respectively; According to formula (4), the initial optimal solution is obtained : set up , , ; Take the limit: , , ; （4） in, is the identity matrix, Indicates The control gain matrix of the iteration is, Indicates the adjustment matrix during iteration Parameters, Indicates The positive solution of the iteration is ; for Other situations: （5） in For the matrix The characteristic value of For the matrix The real part of the eigenvalue, A matrix for adjusting and greater than real number; For each matrix satisfy: Thereby constructing an offline policy learning algorithm.

5. The adaptive cruise control method based on online adaptive learning according to claim 3, characterized in that: Constructing a planning performance function to constrain the offline policy learning algorithm, and processing the offline policy learning algorithm based on the planning performance function to determine an initial state of the online policy learning algorithm, specifically comprising: Construct a smooth function with positive decreasing properties , the smoothing function The formula is: （7） in, , ,satisfy conditions, and As the lower and upper bounds respectively, limit the state of the adaptive cruise control system , is a positive constant greater than 0; based on and Limiting Adaptive Cruise Control System Status , to obtain the transformed adaptive cruise control system state .

6. The adaptive cruise control method based on online adaptive learning according to claim 2, characterized in that: Build an online policy learning algorithm, including: Based on formula (2) and the transformed adaptive cruise control system state ,definition ,definition , define the cost function as ; Two exploration signals used to collect data and Incorporating them into the control input and external disturbance respectively, formula (1) can be rewritten as: （8） Derivative the cost function: (9) in is the cost function of the adaptive cruise control system, which represents the target performance to be achieved by the adaptive cruise control system. Indicates that the adaptive cruise control system is in state , control input and interference input The instantaneous cost of the joint action; To collect data for the adaptive cruise control system, the following matrix is ​​defined: （10） in, ; Define column full rank According to formula (11), we get and ; （11） in, Indicates that it contains Equation Solution The estimated value vector, and They represent the control gain matrices and the interference gain matrix The vectorized form of definition is a positive constant greater than 0, and ,but: （12） in, They are respectively the status of the adaptive cruise control system , control input and interference input The dimension of Based on the above steps, a preliminary online strategy learning algorithm is constructed; wherein, the preliminary online strategy learning algorithm needs to determine the initial matrix ; Remove the initial matrix , to rewrite formula (1) to determine the online policy learning algorithm.

7. The adaptive cruise control method based on online adaptive learning according to claim 6, characterized in that: Formula (1) is rewritten to determine the online policy learning algorithm, which includes: Rewrite formula (1) to generate formula (13): （13） in, , The state matrix of the adaptive cruise control system is adjusted to describe the dynamic changes of the system state; definition and , derive and rewrite formula (9) to obtain formula (14): （14） Through the matrix Solving and Include , and Information problems, using replace , then select , to obtain formula (15): （15） in, , To adjust the weight matrix A positive number, and Respectively represent the maximum eigenvalue and minimum eigenvalue of the matrix; Based on the above steps, the adaptive cruise control data is organized using metrics and the processed data is imported into the matrix tool; , , and After normalization, we get formula (16): （16） By column full rank matrix and , solve : （17） Define a column-full rank matrix: , , , to calculate , and : （18） Based on the above steps, an online policy learning algorithm is obtained.

8. The adaptive cruise control method based on online adaptive learning according to claim 1, characterized in that: Solving the H-infinity control problem based on the online strategy learning algorithm to determine the optimal control strategy specifically includes: Initialize the online policy learning algorithm to , , and ; Repeat control input and interference Collecting data from adaptive cruise control systems ; when satisfy , then according to formula (17) ; According to formula (18), , and ; choose Satisfy the conditions of formula (15) and let Until and ; set up , , and ; According to formula (3), we can solve ; set up , , collect adaptive cruise control system information until formula (12) is satisfied; According to formula (11), we can solve , and ,until and When returned , , , to determine the optimal control strategy.

9. An adaptive cruise control device based on online adaptive learning, characterized in that: The device comprises: at least one processor; and, a memory communicatively coupled to the at least one processor; The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to: Construct the mathematical model of adaptive cruise control system and define the H-infinity control problem of adaptive cruise control system; Defining the state of the adaptive cruise control system based on a preset performance function algorithm and constructing an online strategy learning algorithm; The H-infinity control problem is solved based on the online strategy learning algorithm to determine the optimal control strategy.

10. A non-volatile computer storage medium for adaptive cruise control based on online adaptive learning, storing computer executable instructions, characterized in that: The computer executable instructions are configured to: Construct the mathematical model of adaptive cruise control system and define the H-infinity control problem of adaptive cruise control system; Defining the state of the adaptive cruise control system based on a preset performance function algorithm and constructing an online strategy learning algorithm; The H-infinity control problem is solved based on the online strategy learning algorithm to determine the optimal control strategy.

Images