A Finite-Time Dynamic Surface Dead-Zone-Resistant Tracking Control Method for Unmanned Vehicles

Abstract

This invention discloses a finite-time dynamic surface anti-dead-zone tracking control method for unmanned vehicles, belonging to the field of lane trajectory tracking control for unmanned vehicles. The method includes: constructing a kinematic model of the unmanned vehicle; constructing a front wheel steering angle dead-zone adaptive compensation model, and compensating for asymmetric dead-zone characteristics based on adaptive estimation parameter vectors and activation function vectors, replacing the front wheel steering angle signal in the unmanned vehicle kinematic model with the output of the front wheel steering angle dead-zone adaptive compensation model to form a decoupled unmanned vehicle kinematic model; obtaining a finite-time auxiliary control signal based on the first state variable in the decoupled kinematic model; processing the finite-time auxiliary control signal using a first-order filter to obtain a reference signal for the second state variable; and designing an ideal front wheel steering angle control signal based on the second state variable, the reference signal, and the approximation results for uncertain disturbances. This invention significantly improves the dead-zone compensation capability and dynamic characteristics of the controller.

Description

Technical Field

[0001] This invention belongs to the field of lane line trajectory tracking and control technology for unmanned vehicles, and particularly relates to a finite-time dynamic surface anti-dead-zone tracking and control method for unmanned vehicles. Background Technology

[0002] With the rapid development of autonomous driving technology, accurate lane tracking has become a core requirement for the safe and reliable operation of autonomous vehicles. Existing control methods are mostly based on the assumption of ideal actuators, failing to fully consider the asymmetric dead zone nonlinearity commonly found in real vehicle steering systems. This characteristic significantly reduces steering accuracy and can even lead to system instability. While traditional dynamic surface control and other methods address nonlinearity to some extent, they typically suffer from insufficient dead zone compensation and slow dynamic response convergence, making it difficult to achieve robust and accurate trajectory tracking within a limited time. Especially when the dead zone characteristics are unknown or time-varying, existing control strategies often lack effective adaptive compensation mechanisms, leading to increased tracking errors and severely limiting the control performance of autonomous vehicles in complex dynamic environments. Therefore, a tracking control algorithm capable of handling asymmetric dead zones online while ensuring fast convergence and strong robustness is urgently needed. Summary of the Invention

[0003] To address the aforementioned technical problems, this invention proposes a finite-time dynamic surface anti-dead-zone tracking control method for unmanned vehicles, thereby resolving the issues present in the prior art.

[0004] To achieve the above objectives, the present invention provides a finite-time dynamic surface anti-dead-zone tracking control method for unmanned vehicles, comprising:

[0005] Construct a kinematic model of the unmanned vehicle that includes mapping error, uncertain disturbances, and front wheel steering angle signal affected by asymmetric dead zone characteristics;

[0006] An adaptive compensation model for the front wheel steering angle dead zone is constructed. The adaptive compensation model for the front wheel steering angle dead zone compensates for the asymmetric dead zone characteristics based on the adaptive estimation parameter vector and the activation function vector. The front wheel steering angle signal in the kinematic model of the unmanned vehicle is replaced with the output of the adaptive compensation model for the front wheel steering angle dead zone, forming a decoupled kinematic model of the unmanned vehicle based on the ideal front wheel steering angle signal.

[0007] The finite-time auxiliary control signal is obtained based on the first state variable in the decoupled unmanned vehicle kinematic model.

[0008] The finite-time auxiliary control signal is processed using a first-order filter to obtain a reference signal for the second state variable;

[0009] Based on the second state variable, the reference signal, and the approximation result of the neural network approximator on the uncertain disturbance, an ideal front wheel steering angle control signal is designed.

[0010] Optionally, the process of constructing a kinematic model for an autonomous vehicle includes:

[0011] The lateral error is calculated based on the lateral position of the autonomous vehicle and the reference lateral position.

[0012] The heading error is calculated based on the actual yaw angle and the reference yaw angle of the unmanned vehicle.

[0013] The mapping error is calculated based on the lateral error, the heading error, and the pre-aiming distance;

[0014] The kinematic model of the unmanned vehicle is constructed based on the mapping error, the derivative of the mapping error, the uncertain disturbance function, the control gain coefficient, and the front wheel steering angle signal affected by the asymmetric dead zone characteristics.

[0015] Optionally, the process of constructing the front wheel steering angle dead zone adaptive compensation model includes:

[0016] Construct an adaptive estimation parameter vector based on the dead zone parameter of the front wheel steering angle;

[0017] Construct an activation function vector based on the activation characteristics of the front wheel steering dead zone;

[0018] The ideal front wheel steering angle control signal is processed by using a linear combination of the adaptive estimation parameter vector and the activation function vector to obtain the output of the front wheel steering angle dead zone adaptive compensation model.

[0019] Optionally, the expression for the front wheel steering angle dead zone adaptive compensation model is:

[0020] ;

[0021] In the formula, This is the output of the front wheel steering angle dead zone adaptive compensation model. Represents the adaptive parameter estimation vector; express The transpose of ; This represents the activation function vector.

[0022] Optionally, the expression for the decoupled autonomous vehicle kinematic model is:

[0023] ;

[0024] In the formula, express The first derivative, Indicates mapping error. The first derivative represents the mapping error. express The first derivative, Indicates heading error. Represents velocity in the X direction. Represents the velocity in the Y direction. This represents the control gain coefficient. This represents the output of the front wheel steering dead zone adaptive compensation model. This represents an uncertain perturbation function.

[0025] Optionally, the expression for calculating the finite-time auxiliary control signal is:

[0026] ;

[0027] In the formula, Indicates the first control parameter. This represents the finite-time gain parameter. This indicates a finite-time auxiliary control signal. This indicates the mapping error.

[0028] Optionally, the expression for calculating the reference signal of the second state variable is:

[0029] ;

[0030] In the formula, This represents the filter gain parameter. This indicates the reference signal that the second state variable of the filtered output needs to track. express The first derivative, This indicates a finite-time auxiliary control signal.

[0031] Optionally, the process of designing an ideal front wheel steering angle control signal based on the second state variable, the reference signal, and the approximation result of the neural network approximator on the uncertain disturbance includes:

[0032] The second error variable is obtained based on the difference between the second state variable and the reference signal;

[0033] Based on the second error variable, the approximation result of the neural network approximator, and the control gain coefficient of the decoupled unmanned vehicle kinematic model, the ideal front wheel steering angle control signal is calculated.

[0034] The expression for calculating the ideal front wheel steering angle control signal is as follows:

[0035] ;

[0036] In the formula, This represents the ideal front wheel steering angle control signal. This indicates the second control parameter. This represents the input vector of the neural network. The first derivative represents the mapping error. This represents the finite-time gain parameter. This indicates the result of the neural network approximation. This represents the control gain coefficient.

[0037] The present invention also provides a computer, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method described thereon.

[0038] The present invention also provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described thereon.

[0039] Compared with the prior art, the present invention has the following advantages and technical effects:

[0040] This invention effectively addresses the asymmetric dead zone nonlinearity of autonomous vehicle steering systems by constructing an accurate kinematic model that integrates dead zone characteristic mapping errors and combining an adaptive dead zone compensation mechanism with finite-time dynamic surface control technology. This method can compensate for control signal distortion caused by dead zone in real-time online, transforming the actual nonlinear system into a decoupled model based on ideal front wheel steering angle signals, thereby fundamentally eliminating the interference of dead zone on system accuracy and stability. Simultaneously, the finite-time convergence characteristic ensures rapid convergence of tracking errors, significantly improving the system's transient response speed; the application of dynamic surface technology simplifies the controller structure and avoids the "computational explosion" problem in traditional methods. Ultimately, this enables autonomous vehicles to achieve high-precision and rapid tracking of preset navigation routes even under steering dead zone constraints, significantly enhancing the robustness and engineering practicality of the control system. Attached Figure Description

[0041] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0042] Figure 1 This is a flowchart of a finite-time dynamic surface dead-zone tracking control method for unmanned vehicles according to an embodiment of the present invention;

[0043] Figure 2 This is a diagram showing the change in front wheel steering angle according to an embodiment of the present invention;

[0044] Figure 3 This is a comparison chart of the lateral tracking performance of embodiments of the present invention;

[0045] Figure 4 This is a comparison chart of yaw angle tracking performance in embodiments of the present invention. Detailed Implementation

[0046] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0047] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0048] Example 1.

[0049] This invention significantly improves the dead zone compensation capability and dynamic characteristics of the controller by constructing an accurate kinematic model that integrates dead zone characteristic mapping error and designing a dynamic surface control method for unmanned vehicles that combines adaptive dead zone compensation and finite-time stability.

[0050] like Figure 1 As shown, this embodiment provides a finite-time dynamic surface anti-dead-zone tracking control method for unmanned vehicles, including the following steps:

[0051] Step 1: Construct a kinematic model of the error mapping of the asymmetric dead zone characteristics of the autonomous vehicle;

[0052] Step 2: Use the front wheel steering angle dead zone adaptive compensation model to handle the asymmetric dead zone characteristics, thereby forming a decoupled unmanned vehicle kinematic model based on the ideal front wheel steering angle control signal;

[0053] Step 3: Design the ideal front wheel steering angle control signal using finite-time dynamic surface technology.

[0054] The specific method for constructing an error kinematic model of the asymmetric dead zone characteristics of an autonomous vehicle is as follows:

[0055] While driving along a given navigation path, the autonomous vehicle needs to simultaneously control its lateral error. and heading error The lateral error is defined as the difference between the actual position y in the Y direction and the reference position y in the Y direction. The difference: Similarly, heading error is defined as the actual yaw angle. Compared with reference yaw angle The difference: However, in order for a single front wheel steering angle control signal to simultaneously control lateral error... and heading error If the error approaches zero, then these two types of errors need to be integrated, thus defining a mapping error as follows:

[0056] (1)

[0057] in, This represents the aiming distance along the vehicle speed direction. It is an adjustable design parameter, typically taken as 1 to 2 times the wheelbase of the autonomous vehicle. Combining the definition formula (1) for mapping error, the kinematic model of the asymmetric dead zone characteristic mapping error of a certain type of autonomous vehicle can be described as follows:

[0058] (2)

[0059] in, , , , , These represent the mapping error, the first derivative of the mapping error, the uncertain disturbance function, the control gain coefficient, and the front wheel steering angle control signal affected by the asymmetric dead zone characteristics, respectively. for The first derivative, for The first derivative. Control gain coefficient. It mainly relates to some inherent parameters of the autonomous vehicle itself, and can be defined in the following form:

[0060] (3)

[0061] in, Indicates the front wheelbase. The coefficient of friction of the road surface. The steering stiffness of the left front wheel. The steering stiffness of the right front wheel. Represents the yaw moment of inertia. Front wheel steering angle control signal affected by asymmetric dead zone characteristics. Defined in the following form:

[0062] (4)

[0063] in, For ideal front wheel steering angle control signal, , , , These represent the right slope, left slope, right boundary, and left boundary of the dead zone model, respectively. Because the front wheel steering angle execution structure has a dead zone, the designed nominal front wheel steering angle control signal... The range of variation is in In between, The output will tend to zero. Furthermore, The absolute value is not necessarily equal to The absolute value of the actuator dead zone considered in this invention has asymmetric characteristics.

[0064] An adaptive compensation model for front wheel steering angle dead zone is used to handle asymmetric dead zone characteristics, thereby forming a decoupled kinematic model of the unmanned vehicle based on the ideal front wheel steering angle control signal. The specific method is as follows:

[0065] First, the front wheel steering angle control signal is affected by the asymmetric dead zone characteristics. The dead zone of the front wheel steering angle is mainly affected by four key parameters: right slope, left slope, right boundary, and left boundary. However, in actual control environments, these four key parameters are often impossible to measure accurately. Therefore, an adaptive estimation technique is needed to estimate the true values ​​of these four parameters in real time, which requires the design of an adaptive compensation model for the dead zone of the front wheel steering angle. :

[0066] (5)

[0067] in, Represents the adaptive parameter estimation vector; express The transpose of ; Represents the activation function vector. Adaptive parameter estimation vector. Its main function is to estimate four key parameters: right slope, left slope, right boundary, and left boundary. Therefore, it includes four estimated variables, namely:

[0068] = (6)

[0069] in, , , , Let represent the right slope estimate, left slope estimate, right boundary estimate, and left boundary estimate, respectively. The specific values ​​of these four estimated variables are estimated in real time using the following dynamic differential equation:

[0070] (7)

[0071] in, , , , They represent , , and The derivative; , , , , All of these represent design parameters.

[0072] Similarly, estimate the member function vector. It is also defined as a four-dimensional vector:

[0073] (8)

[0074] in, and Let represent the right dead zone activation function and the left dead zone activation function, respectively. This invention selects a class of continuous functions with exponential decay characteristics as the activation function, thus... and It was further designed as:

[0075] (9)

[0076] in, and All of these represent exponential decay design parameters. Represents an exponential function. Current wheel angle dead zone adaptive compensation model. During convergence, it can estimate the front wheel steering angle control signal affected by asymmetric dead zone characteristics in real time. Therefore, the front wheel steering angle control signal design algorithm proposed in this invention can have a certain compensation estimation capability even when there is a dead zone in the actuator that drives the front wheel steering angle.

[0077] Then, in formula (2) Replace with This allows us to obtain a decoupled kinematic model of the autonomous vehicle based on the ideal front wheel steering angle control signal:

[0078] (10)

[0079] The ideal front wheel steering angle control signal is designed using finite-time dynamic surface technology, and the specific method is as follows;

[0080] Front wheel steering dead zone adaptive compensation model The expression reveals that its adaptive parameter estimation vector The relevant member variables have been estimated in real time using an adaptive update law, and the member function vector... The activation function in the design has also been further refined. Therefore, the finite-time dynamic surface technique will be used to further construct the ideal front wheel steering angle control signal. This completes the design of the entire finite-time dynamic surface anti-dead-zone tracking control method.

[0081] First, it is necessary to provide the first-order subsystem. The first state variable in Find the reference signal, the first state variable That is, the mapping error defined by formula (1). However, when the mapping error approaches zero, it indicates that the actual position y in the Y direction approaches the reference position in the Y direction. And the actual yaw angle tending towards reference yaw angle This means that the autonomous vehicle is gradually approaching the preset navigation path. Therefore, from the above analysis, we can know the state variables... The reference signal to be tracked can be set to zero. Therefore, the first state variable can be directly set to zero. Consider it as the first error variable corresponding to the first-order subsystem. Then, regarding the first state variable... The following finite-time auxiliary control signal can be constructed. :

[0082] (11)

[0083] in, Indicates the first control parameter. This represents the finite-time gain parameter, and its range of values ​​is... Designed for the first-order subsystem Enable state variables It gradually approaches zero. It can be observed that when... Then, equation (11) above becomes a common auxiliary control signal. . This can be considered a proportional control signal, which lacks finite-time gain characteristics and therefore cannot improve the response speed of the control signal. However, the finite-time auxiliary control signal designed in this invention can improve the response speed by reducing the finite-time gain parameter. The value of is used to further improve the dynamic response characteristics of the control algorithm.

[0084] Then, in order to give the second-order subsystem The second state variable in Finding a suitable reference signal, the second state variable That is, the first derivative of the mapping error defined by formula (1) can be used to introduce a first-order filter:

[0085] (12)

[0086] in, This represents the filter gain parameter. Indicates the filtered output signal. express The first derivative. This invention will... Treat as a second state variable The reference signal that needs to be tracked. Therefore, the second error variable is constructed as follows for the second-order subsystem: :

[0087] (13)

[0088] Because the second-order subsystem has an uncertain perturbation function Therefore, in designing an ideal front wheel steering angle control signal Previously, the following neural network approximator also needed to be designed. :

[0089] (14)

[0090] in, This represents the weighted parameter estimation variable. Represents the activation function vector. express transpose, This represents the constant design parameters. The input vector of the neural network is further defined as follows:

[0091] (15)

[0092] in, Dimensions and They have the same dimensions. Therefore, The design is unfolded into the following form:

[0093] (16)

[0094] in, , , indicating the first An activation member function, designed in the following form:

[0095] (17)

[0096] in, Indicates the width at the center position. This represents a 4xM matrix showing the center positions of nodes. The specific value of parameter M is the same as the number of nodes in the neural network.

[0097] Next, the ideal front wheel steering angle control signal can be designed as follows. :

[0098] (18)

[0099] in, This indicates the second control parameter. The weight parameter estimation variables involved in the process The following dynamic differential equation with weight parameters is used. Perform real-time estimation:

[0100] (19)

[0101] in, and These represent the first design parameter and the second design parameter, respectively.

[0102] First, the control method of this invention effectively solves the adverse effects of asymmetric dead zone nonlinearity in the actuator by constructing an error kinematic model mapping the asymmetric dead zone characteristics of the unmanned vehicle and introducing an adaptive compensation model for the front wheel steering angle dead zone. This design can compensate for the control signal distortion caused by the dead zone characteristics online in real time, transforming the actual nonlinear system into a decoupled unmanned vehicle kinematic model based on the ideal front wheel steering angle control signal, thereby eliminating the interference of dead zone on the system's accuracy and stability from a mechanistic perspective. This active anti-dead zone strategy does not rely on precise prior knowledge of dead zone parameters, significantly enhancing the robustness of the control system to uncertainties in the actuator and ensuring the basic control performance of the unmanned vehicle under the constraint of hardware nonlinearity and asymmetric dead zone.

[0103] Secondly, this invention innovatively combines finite-time convergence theory with dynamic surface control technology to design an ideal front wheel steering angle control signal. The finite-time control law ensures that the tracking error system can quickly converge to the neighborhood of the error equilibrium point in a short time, significantly improving the transient response speed and convergence performance of the autonomous vehicle in tracking the navigation route. Simultaneously, the introduction of dynamic surface technology effectively avoids the inherent "computational explosion" problem in traditional backstepping designs, simplifies the controller structure, and improves the algorithm's feasibility and engineering practicality. The synergistic effect of these two technologies enables the autonomous vehicle system to not only possess high-precision, fast-convergence tracking capabilities for preset navigation routes, but also maintain superior dynamic quality and steady-state accuracy even under practical constraints such as the front wheel steering angle dead zone.

[0104] Example 2.

[0105] A finite-time dynamic surface tracking control method for unmanned vehicles, comprising the following steps:

[0106] (1) Construct a kinematic model of the error mapping of the asymmetric dead zone characteristics of unmanned vehicles;

[0107] (2) Construct an unmanned vehicle asymmetric dead zone characteristic mapping error kinematic model and use the front wheel steering angle dead zone adaptive compensation model to process the asymmetric dead zone characteristics, thereby forming a decoupled unmanned vehicle kinematic model based on the ideal front wheel steering angle control signal;

[0108] (3) Design an ideal front wheel steering angle control signal using finite-time dynamic surface technology;

[0109] (4) Data analysis of the actual application results of the controller.

[0110] 1. Construct a kinematic model of the error mapping of the asymmetric dead zone characteristics of unmanned vehicles.

[0111] First, based on the horizontal error and heading error The mapping error is defined as follows:

[0112] (20)

[0113] Among them, the pre-aiming distance along the vehicle speed direction Set to 3.1. Combining the definition of mapping error (20), the kinematic model of the asymmetric dead zone characteristic mapping error of a certain type of autonomous vehicle can be described as...

[0114] (twenty one)

[0115] in, , , , , These represent the mapping error, the first derivative of the mapping error, the uncertain disturbance function, the control gain coefficient, and the front wheel steering angle control signal affected by the asymmetric dead zone characteristics, respectively. for The first derivative, for The first derivative. Control gain coefficient. It mainly relates to some inherent parameters of the autonomous vehicle itself, and can be defined in the following form:

[0116] (twenty two)

[0117] Among them, front wheelbase Steering stiffness of the left front wheel Steering stiffness of the right front wheel Yaw moment of inertia This embodiment simulates the operating conditions of an unmanned vehicle on an icy or snowy road surface. Therefore, the road surface friction coefficient is relatively low and is set to... .

[0118] Then, the front wheel steering angle control signal affected by the asymmetric dead zone characteristics Defined in the following form:

[0119] (twenty three)

[0120] in, This is the ideal front wheel steering angle control signal. In this embodiment, the dead zone key parameter value of the front wheel steering angle actuator is the right slope. Left slope Right boundary and left boundary It can be observed that... The absolute value (0.5) is not equal to The absolute value of is 0.65. Therefore, the dead zone of the front wheel steering actuator involved in this embodiment has asymmetric characteristics.

[0121] 2. The asymmetric dead zone characteristics are handled by using the front wheel steering angle dead zone adaptive compensation model, thereby forming a decoupled unmanned vehicle kinematic model based on the ideal front wheel steering angle control signal.

[0122] First, an adaptive estimation technique is used to estimate the true values ​​of four key dead zone characteristic parameters—right slope, left slope, right boundary, and left boundary—in real time, thereby designing an adaptive compensation model for front wheel steering angle dead zone. :

[0123] (twenty four)

[0124] in, Represents the adaptive parameter estimation vector; express The transpose of ; Represents the activation function vector. Adaptive parameter estimation vector. Its main function is to estimate four key parameters: right slope, left slope, right boundary, and left boundary. Therefore, it includes four estimated variables, namely:

[0125] = (25)

[0126] in , , , Let represent the right slope estimate, left slope estimate, right boundary estimate, and left boundary estimate, respectively. The specific values ​​of these four estimated variables are estimated in real time using the following dynamic differential equation:

[0127] (26)

[0128] in, , , , They represent , , and The derivative; design parameters , , , , .

[0129] Similarly, estimate the member function vector. It is also defined as a four-dimensional vector:

[0130] (27)

[0131] in, and Let represent the right dead zone activation function and the left dead zone activation function, respectively. In this embodiment, a class of continuous functions with exponential decay characteristics is selected as the activation function. and It was further designed as:

[0132] (28)

[0133] Among them, the exponential decay design parameters and , This represents an exponential function.

[0134] Then, in formula (2) Replace with This allows us to obtain a decoupled kinematic model of the autonomous vehicle based on the ideal front wheel steering angle control signal:

[0135] (29)

[0136] 3. The ideal front wheel steering angle control signal is designed using finite-time dynamic surface technology.

[0137] First, the state variables can be directly... This is considered as the first error variable corresponding to the first-order subsystem. Therefore, regarding the first error variable... The following finite-time auxiliary control signal can be constructed. :

[0138] (30)

[0139] Among them, control parameters Finite-time gain parameter .

[0140] Then, in order to give the second-order subsystem State variables in To find a suitable reference signal, a first-order filter needs to be introduced:

[0141] (31)

[0142] Among them, the filter gain parameter , Indicates the filtered output signal. express The first derivative. This embodiment will... Treat as state variables The reference signal that needs to be tracked. Therefore, the second error variable is constructed as follows for the second-order subsystem: :

[0143] (32)

[0144] Because the second-order subsystem has an uncertain perturbation function Therefore, in designing an ideal front wheel steering angle control signal Previously, the following neural network approximator also needed to be designed. :

[0145] (33)

[0146] in, This represents the weighted parameter estimation variable. Represents the activation function vector. express Transpose, often designed parameters . The input vector of the neural network is further defined as follows:

[0147] (34)

[0148] in, Dimensions and They have the same dimensions. Therefore, The design is unfolded into the following form:

[0149] (35)

[0150] in, , , indicating the first An activation member function, designed in the following form:

[0151] (36)

[0152] Among them, the width of the center position In this embodiment, the number of nodes in the neural network is set to 10. Therefore, the center position matrix... It is a 4x10 matrix with the following specific parameters:

[0153] (37)

[0154] Next, the ideal front wheel steering angle control signal can be designed as follows. :

[0155] (38)

[0156] Among them, control parameters , The weight parameter estimation variables involved in the process The following dynamic differential equation with weight parameters is used. Perform real-time estimation:

[0157] (39)

[0158] Among them, design parameters and .

[0159] 4. Data analysis of actual application results of the controller.

[0160] First, the control algorithm and its control parameters of this invention are deployed into the autonomous driving control module of the unmanned vehicle, and it is made to track the following reference trajectory:

[0161] (40)

[0162] in, This indicates the initial speed of the driverless car; Represents a time variable; This represents the coordinate value of the autonomous vehicle in the X-axis direction of the global coordinate system (i.e., its longitudinal position). Represents the hyperbolic tangent function. This represents the inverse cosine function. The vertical position is calculated using the following formula in this embodiment. :

[0163] (41)

[0164] Among them, the initial velocity Even with an asymmetric dead zone in the front wheel steering mechanism, the front wheel steering angle control algorithm of this invention has achieved satisfactory practical application results, such as... Figure 2 , Figure 3 and Figure 4 As shown. Figure 2 The curve showing the change of the front wheel steering angle over time t is displayed. This represents the change in the front wheel steering angle over time t. On one hand, it can be observed that the change in the front wheel steering angle shrinks to within the dead zone. The chattering phenomenon begins at this time, such as Figure 2 The dead zone time falls within the 1.6-2.9 range, which significantly impacts the lifespan of the actuator. However, during this period, the dead zone compensation model is using relevant state information from sensor feedback to adaptively correct and train its weight parameters, thereby improving the accuracy of its dead zone parameter estimation. Once the dead zone compensation model accurately estimates and compensates for the key dead zone parameters, the front wheel steering angle no longer exhibits vibration and tends towards a stable and smooth change, such as... Figure 2 The change curve after 3.5 seconds is shown. Furthermore, by adjusting the finite-time gain parameter, the dynamic response speed of the controller is also improved. For example, the lateral displacement of the autonomous vehicle... It tracked the Y-direction reference position within 0.26 seconds. Furthermore, the tracking error has been reduced to within 0.011m, such as Figure 3 As shown; the yaw angle of the unmanned vehicle. It tracked the reference yaw angle within 0.3 seconds. And the tracking error is controlled within 0.01 degrees, such as Figure 4 As shown.

[0165] This embodiment also provides a computer, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method described thereon.

[0166] This embodiment also provides a storage medium on which a computer program is stored, which, when executed by a processor, implements the method described thereon.

[0167] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A finite-time dynamic surface anti-dead-zone tracking control method for unmanned vehicles, characterized in that, Includes the following steps: Construct a kinematic model of the unmanned vehicle that includes mapping error, uncertain disturbances, and front wheel steering angle signal affected by asymmetric dead zone characteristics; An adaptive compensation model for the front wheel steering angle dead zone is constructed. The adaptive compensation model for the front wheel steering angle dead zone compensates for the asymmetric dead zone characteristics based on the adaptive estimation parameter vector and the activation function vector. The front wheel steering angle signal in the kinematic model of the unmanned vehicle is replaced with the output of the adaptive compensation model for the front wheel steering angle dead zone, forming a decoupled kinematic model of the unmanned vehicle based on the ideal front wheel steering angle signal. The finite-time auxiliary control signal is obtained based on the first state variable in the decoupled unmanned vehicle kinematic model. The finite-time auxiliary control signal is processed using a first-order filter to obtain a reference signal for the second state variable; Based on the second state variable, the reference signal, and the approximation result of the neural network approximator on the uncertain disturbance, an ideal front wheel steering angle control signal is designed.

2. The finite-time dynamic surface anti-dead-zone tracking control method for unmanned vehicles according to claim 1, characterized in that, The process of constructing a kinematic model for an autonomous vehicle includes: The lateral error is calculated based on the lateral position of the autonomous vehicle and the reference lateral position. The heading error is calculated based on the actual yaw angle and the reference yaw angle of the unmanned vehicle. The mapping error is calculated based on the lateral error, the heading error, and the pre-aiming distance; The kinematic model of the unmanned vehicle is constructed based on the mapping error, the derivative of the mapping error, the uncertain disturbance function, the control gain coefficient, and the front wheel steering angle signal affected by the asymmetric dead zone characteristics.

3. The finite-time dynamic surface anti-dead-zone tracking control method for unmanned vehicles according to claim 1, characterized in that, The process of constructing the adaptive compensation model for front wheel steering dead zone includes: Construct an adaptive estimation parameter vector based on the dead zone parameter of the front wheel steering angle; Construct an activation function vector based on the activation characteristics of the front wheel steering dead zone; The ideal front wheel steering angle control signal is processed by using a linear combination of the adaptive estimation parameter vector and the activation function vector to obtain the output of the front wheel steering angle dead zone adaptive compensation model.

4. The finite-time dynamic surface anti-dead-zone tracking control method for unmanned vehicles according to claim 3, characterized in that, The expression for the adaptive compensation model for the front wheel steering dead zone is: ； In the formula, This is the output of the front wheel steering angle dead zone adaptive compensation model. Represents the adaptive parameter estimation vector; express The transpose of ; This represents the activation function vector.

5. The finite-time dynamic surface anti-dead-zone tracking control method for unmanned vehicles according to claim 1, characterized in that, The expression for the decoupled autonomous vehicle kinematic model is as follows: ； In the formula, express The first derivative, Indicates mapping error. The first derivative represents the mapping error. express The first derivative, Indicates heading error. Represents velocity in the X direction. Represents the velocity in the Y direction. This represents the control gain coefficient. This represents the output of the front wheel steering dead zone adaptive compensation model. This represents an uncertain perturbation function.

6. The finite-time dynamic surface anti-dead-zone tracking control method for unmanned vehicles according to claim 1, characterized in that, The expression for calculating the finite-time auxiliary control signal is as follows: ； In the formula, Indicates the first control parameter. This represents the finite-time gain parameter. This indicates a finite-time auxiliary control signal. This indicates the mapping error.

7. The finite-time dynamic surface anti-dead-zone tracking control method for unmanned vehicles according to claim 1, characterized in that, The expression for calculating the reference signal of the second state variable is: ； In the formula, This represents the filter gain parameter. This indicates the reference signal that the second state variable of the filtered output needs to track. express The first derivative, This indicates a finite-time auxiliary control signal.

8. The finite-time dynamic surface anti-dead-zone tracking control method for unmanned vehicles according to claim 1, characterized in that, The process of designing the ideal front wheel steering angle control signal based on the second state variable, the reference signal, and the approximation result of the neural network approximator for the uncertain disturbance includes: The second error variable is obtained based on the difference between the second state variable and the reference signal; Based on the second error variable, the approximation result of the neural network approximator, and the control gain coefficient of the decoupled unmanned vehicle kinematic model, the ideal front wheel steering angle control signal is calculated. The expression for calculating the ideal front wheel steering angle control signal is as follows: ； In the formula, This represents the ideal front wheel steering angle control signal. This indicates the second control parameter. This represents the input vector of the neural network. The first derivative represents the mapping error. This represents the finite-time gain parameter. This indicates the result of the neural network approximation. This represents the control gain coefficient.

9. A computer comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method as described in claim 1.

10. A storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in claim 1.

Images