Model-free data-driven time-varying performance control method for unmanned surface vehicle

Abstract

The application discloses a kind of model-free unmanned ship data-driven specified time performance control method, including error conversion function based on the specified time performance function of construction;Based on the unmanned ship data-driven neural predictor of construction, to realize the estimation of model uncertainty and unknown control input gain, obtain the unmanned ship dynamics model;Based on error conversion function, according to the unmanned ship dynamics model, model-free data-driven virtual control law is constructed, to construct data-driven controller;According to the switching dynamic event triggering mechanism of data-driven controller design for switching fixed threshold triggering mechanism and dynamic event triggering mechanism, and realize the unmanned ship data-driven specified time performance control according to switching dynamic event triggering mechanism.Solve the problem that existing unmanned ship control process cannot effectively improve the control performance of underactuated unmanned ship through data-driven, and ensure that the specified trajectory precision difference is reached within the preset time, in addition, the existing method based on neural network mostly relies on a large number of sample training fixed model, when facing new task or environment change, there is the problem of limited adaptability of training model.

Description

Technical Field

[0001] This invention relates to the field of unmanned vessel control technology, and in particular to a time-based performance control method for model-free unmanned vessels driven by data. Background Technology

[0002] In recent years, unmanned surface vessels (USVs) have seen numerous applications in the marine field, including ocean monitoring, target search, and data collection. With the advancement of ocean exploration, the increasing complexity of operations places higher demands on the control precision of USV systems. The complex cascaded structure and harsh navigation environment of USVs make it difficult to obtain accurate mathematical models. Furthermore, communication bandwidth is limited during USV operations at sea. Therefore, traditional model-based control methods are insufficient to meet the demands in terms of both accuracy and resource utilization.

[0003] Currently, time-triggered and event-triggered mechanisms are widely studied to save communication resources. In time-triggered mechanisms, signal transmission is determined by the sampling time. In event-triggered mechanisms, signal transmission is determined by events designed based on system requirements. Event triggering parameters are divided into static and dynamic types. Generally, dynamic event triggering mechanisms have longer trigger intervals compared to static triggering mechanisms, and therefore receive more attention. For high-precision control systems, sudden changes in control signals can lead to unnecessary updates to the triggering mechanism when applying traditional event triggering mechanisms. Switching between dynamic event triggering mechanisms effectively solves this problem by switching between fixed threshold triggering mechanisms and dynamic event triggering mechanisms. However, due to the unknown model characteristics of underactuated unmanned surface vessels (USVs) in actual sea conditions, no research has yet explored the application of switching dynamic event triggering mechanisms in model-less USVs.

[0004] Furthermore, unmanned surface vessels (USVs) frequently encounter interference from internal system dynamics and external environmental factors such as wind, waves, and ocean currents, affecting system performance. To address these uncertainties, methods such as extended state observers, neural networks, and fuzzy logic systems have been widely adopted. Among these, neural networks have attracted significant attention due to their ability to learn and optimize control strategies, with some results showing promising progress in tasks such as trajectory tracking. However, most existing neural network-based methods rely on training fixed models with a large number of samples, resulting in limited adaptability when faced with new tasks or environmental changes. Summary of the Invention

[0005] This invention provides a time-based performance control method for model-free unmanned surface vessels driven by data, in order to overcome the aforementioned technical problems.

[0006] To achieve the above objectives, the technical solution of the present invention is as follows:

[0007] A data-driven, time-based performance control method for model-free unmanned surface vessels includes the following steps:

[0008] S1: Obtain the underactuated unmanned vessel mathematical model containing model uncertainties;

[0009] S2: Define and obtain the tracking error between the actual trajectory and the expected trajectory based on the mathematical model of the unmanned vessel;

[0010] And obtain the error transformation function based on the constructed time performance function;

[0011] S3: Obtain the system dynamics model including model uncertainties and unknown control input gains based on the unmanned ship mathematical model, and use the constructed unmanned ship data-driven neural predictor to estimate the model uncertainties and unknown control input gains to obtain the unmanned ship dynamics model;

[0012] S4: Based on the error transformation function, construct a model-free data-driven virtual control law according to the dynamic model of the unmanned vessel;

[0013] S5: Define the kinematic tracking error variable based on the model-free data-driven virtual control law, and construct a data-driven controller based on the kinematic tracking error variable;

[0014] S6: Based on the data-driven controller design, a switching dynamic event triggering mechanism is used to switch between the fixed threshold triggering mechanism and the dynamic event triggering mechanism, and the unmanned vessel data-driven performance control at specified times is realized based on the switching dynamic event triggering mechanism.

[0015] Furthermore, the expression for the underactuated unmanned surface vessel mathematical model, which includes model uncertainties, obtained in S1 is as follows:

[0016]

[0017]

[0018] In the formula: This indicates the unmanned vessel's x-coordinate, y-coordinate, and bow angle in a preset Earth coordinate system. This indicates the longitudinal velocity, lateral velocity, and angular velocity of the unmanned surface vessel. Indicates the internal model uncertainty term; This indicates unknown disturbances in the environment and , Represents the upper bound of unknown interference; This indicates the control inputs provided by the thrusters and rudders; express The first derivative; express The first derivative; , This indicates the added quality of the model.

[0019] Furthermore, S2 specifically includes the following steps:

[0020] S21: Obtain the relative distance and relative azimuth angle based on the unmanned surface vessel's mathematical model; its expression is as follows:

[0021]

[0022]

[0023] In the formula: Indicates relative distance and , ; These represent the minimum and maximum preset relative distances, respectively. Indicates azimuth; Indicates relative azimuth and and ; Indicates the upper bound of the relative azimuth angle;

[0024] Based on the relative distance and relative azimuth, the tracking error between the actual trajectory and the desired trajectory is defined and obtained, and the expression for the tracking error is:

[0025] ,

[0026] In the formula: Indicates distance tracking error; Represents the expected trajectory distance and ; This indicates the azimuth tracking error. Indicates the desired azimuth angle and ;

[0027] S22: Construct the specified time performance function, the expression of which is:

[0028]

[0029]

[0030]

[0031]

[0032] In the formula: A boundary performance function representing a specified time; express The first derivative; and Indicates a positive design parameter; , , Indicates the amount of intermediate parameters; express The first derivative; Indicates the set time; Indicates the quantity of intermediate parameters and ; and Indicates time design parameters; Indicates constant gain; Indicates the convergence time;

[0033] S23: Based on the constructed time performance function, obtain the constraints on the tracking error;

[0034] And the expression for the constraint condition is:

[0035]

[0036] In the formula: and Indicates the boundary gain coefficient;

[0037] S24: Obtain the error transformation function based on the constraints of the tracking error, wherein the expression of the error transformation function is...

[0038]

[0039]

[0040] In the formula: / Indicates intermediate parameter variables; , , , , , All represent design parameters; Represents the tracking error variable between the virtual ship and the following ship, and is The abbreviated form; This represents the error transformation function.

[0041] Furthermore, S3 specifically includes the following steps:

[0042] S31: Construct a data-driven neural predictor for unmanned ships based on the mathematical model of unmanned ships;

[0043] The unmanned vessel data-driven neural predictor includes a cascaded network unit composed of a neural network module and a filtering module, and a predictor development unit;

[0044] S311: Obtain the system dynamic equations of the mathematical model of the underactuated unmanned surface vessel, including model uncertainties. The expression of the system dynamic equations is as follows:

[0045]

[0046] In the formula: , , This represents the internal model uncertainty term of the unmanned vessel's mathematical model; , Indicates unknown control input gain; , Indicates system control input;

[0047] S312: The system dynamics equations are approximated using a neural network module, and the expression is:

[0048]

[0049]

[0050] In the formula: Indicates the current time step; and Indicates the time step of the delay; Indicates transpose; This indicates the uncertainty of the system; Indicates period; Represents the Gaussian function; and Let represent the weight vector and approximation error of the neural network, respectively, and satisfy . and , and Indicates design constants;

[0051] S313: A filter is constructed using the filtering module to filter the uncertainties in the internal model after approximation processing. Its expression is:

[0052]

[0053] ,

[0054]

[0055] In the formula: This represents an estimate of the gain for an unknown input. Indicates external input; This represents the filtered parameter vector; This represents the velocity derivative after filtering; Indicates a positive design constant; express The first derivative; Represents the filter parameter vector; express The initial value; express The filtered value and ; This represents a parameter used to compensate for system uncertainties. Indicates filter parameters; Represents the input feature vector of a neural network module ; The filter's parameter weight vector and ;

[0056] S314: Obtain the historical dataset driven by unmanned vessel data stored in the preset memory stack;

[0057] And the historical dataset is denoted as ;

[0058] in, and , Indicates the stack length of the memory stack; Indicates time t=k The filter parameter vector at the location; Indicates time t=k The derivative of the filtering velocity at that point; Indicates time t=k Input gain estimate at the location;

[0059] S315: Based on cascaded network units, a data-driven neural predictor for unmanned surface vessels is constructed through a predictor development unit. Its expression is as follows:

[0060]

[0061]

[0062] In the formula: , , Represents design constants and ; express The estimated value and ; express The estimated value; Represents a positive definite coefficient matrix; This represents the adjustment factor for weight updates; Indicates the stack length of the memory stack; Represents the kinematic tracking error variable;

[0063] S316: An unmanned vessel data-driven neural predictor is constructed using reinforcement learning based on historical datasets. Its parameter weights are confirmed to reconstruct the unmanned vessel data-driven neural predictor, which is used to estimate the uncertainties of the model and obtain the dynamic model of the unmanned vessel.

[0064] Furthermore, the expression for the model-free data-driven virtual control law constructed in S4 is:

[0065]

[0066]

[0067]

[0068] In the formula: , , These represent virtual control laws for longitudinal, lateral, and steering directions, respectively. This represents an estimate of the unknown input gain. , Indicates the error transformation scaling factor; express The first derivative; , , , , , , , , Indicate design parameters; express The first derivative; Indicates intermediate parameters and , , ; , , Indicates the design scaling parameters; , , Represents the auxiliary variable used to adjust the dynamic behavior of errors in the control system, and the auxiliary system is... ; , This represents an adaptive parameter used to adjust the response strength of auxiliary system variables to control deviations and system states. Indicates control gain; Indicates intermediate parameters and , , .

[0069] Furthermore, S5 specifically includes the following steps:

[0070] S51: Based on the virtual control law and the unmanned vessel dynamics model, define the kinematic tracking error variable, whose expression is:

[0071]

[0072] In the formula: These represent the tracking error variables in the longitudinal, lateral, and steering directions, respectively.

[0073] S52: Construct a data-driven controller based on the kinematic tracking error variable; its expression is as follows:

[0074]

[0075]

[0076]

[0077]

[0078] In the formula: , These represent the longitudinal and steering direction control inputs, respectively. , The tracking error variables represent the longitudinal and steering directions; , Indicates intermediate variables; , This indicates the desired speed command that guides the system towards an ideal state. , Indicates control gain; , Describe the basis functions; , This represents an estimate of the unknown input gain. , This represents the dynamic threshold parameter.

[0079] Furthermore, the dynamic event triggering mechanism designed in S6 for switching between the fixed threshold triggering mechanism and the dynamic event triggering mechanism has the following expression:

[0080]

[0081]

[0082]

[0083] In the formula: Indicates in The data at any given moment drives the controller's output; Indicates the triggering error of switching dynamic event triggering mechanisms and ; and This indicates a dynamically adjustable threshold parameter; This represents the threshold parameter used to switch between the fixed threshold triggering mechanism FTTM and the dynamic event triggering mechanism DETM; , , Indicate design parameters; and Indicates a dynamically adjustable threshold parameter , The first derivative; Indicates time interval The actual control input within; Indicates the time parameter.

[0084] This invention provides a time-bound performance control method for model-free unmanned surface vessels driven by data, with the following advantages:

[0085] 1. By constructing an unmanned vessel data-driven neural predictor, the model uncertainty and position input gain in the system dynamics model are estimated, thus obtaining the unmanned vessel dynamics model. Based on this, a model-free data-driven virtual control law is designed. This unmanned vessel data-driven neural predictor can adapt to new tasks, model changes, and environmental changes without relying on prior information of model parameters. Unlike traditional deterministic learning methods that are limited to repetitive tasks, this method is more flexible and adaptable.

[0086] 2. The constructed time-defined performance function has shape adaptability and preset time convergence. Its convergence time can be defined by the user and is independent of the system initial conditions and system parameters. Compared with previous work, this time-defined performance function can be applied to unmanned ships without fixed data models, enhancing its adaptability to different environments.

[0087] 3. The designed switching dynamic event triggering mechanism can effectively reduce the communication burden. By dynamically switching between the fixed threshold triggering mechanism and the dynamic event triggering mechanism, communication resources are further saved. The advantages are more obvious, especially when the actuator jumps or changes suddenly during the operation of the controller. Attached Figure Description

[0088] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0089] Figure 1This is a flowchart of the prescribed time performance control method of the present invention applied to data-driven model-free unmanned surface vessels;

[0090] Figure 2 This is a schematic diagram illustrating the performance control triggered by the switching time in this embodiment;

[0091] Figure 3 This is a planar trajectory diagram comparing the method of this embodiment with existing algorithms;

[0092] Figure 4 This is a diagram showing the positional error between the method in this embodiment and existing algorithms;

[0093] Figure 5 This is a diagram showing the angle error between the method in this embodiment and existing algorithms;

[0094] Figure 6 This is a simulation diagram showing the approximation accuracy of the data-driven neural predictor in this example;

[0095] Figure 7 This is a graph showing the estimation effect of the method in this example on unknown input gain;

[0096] Figure 8 This example shows the continuous dynamic adjustment history of the actual control input over time.

[0097] Figure 9 This is a distribution diagram of adjacent trigger time intervals in this example. Detailed Implementation

[0098] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0099] This embodiment provides a time-bound performance control method for model-free unmanned surface vessels (USVs) driven by data, such as... Figure 1 As shown, the specific steps include:

[0100] S1: Obtain the underactuated unmanned vessel mathematical model containing model uncertainties;

[0101] In this embodiment, a mathematical model containing unknown terms in the mathematical model of an unmanned vessel is defined as a model-free model. That is, although the descriptive formula of the mathematical model of the unmanned vessel can be written, it contains uncertain unknown terms, such as internal model uncertain terms and unknown disturbance terms about the environment.

[0102] Specifically, the expression for the underactuated unmanned surface vessel mathematical model, which includes model uncertainties, obtained in S1 is as follows:

[0103]

[0104]

[0105] In the formula: This indicates the unmanned vessel's x-coordinate, y-coordinate, and bow angle in a preset Earth coordinate system. This indicates the longitudinal velocity, lateral velocity, and angular velocity of the unmanned surface vessel. Indicates the internal model uncertainty term; This indicates unknown disturbances in the environment and , Represents the upper bound of unknown interference; This indicates the control inputs provided by the thrusters and rudders; express The first derivative; express The first derivative; , Indicates the added mass of the model;

[0106] S2: Define and obtain the tracking error between the actual trajectory and the expected trajectory based on the mathematical model of the unmanned vessel;

[0107] And based on the constructed time performance function, the error transformation function is obtained, specifically including the following steps:

[0108] S21: In this embodiment, the desired trajectory is obtained through a preset leader ship, and the speed of the leader is defined. and It is bounded, that is and ,in and express and The upper bound;

[0109] The relative distance and relative azimuth angle are obtained from the mathematical model of the unmanned surface vessel, and their expressions are as follows:

[0110]

[0111]

[0112] In the formula: Indicates relative distance; Representing relative azimuth angles, and considering the physical limitations of sensors as well as collision avoidance and connectivity requirements, the relative distance and angle must meet the following conditions: , , and ; These represent the minimum and maximum preset relative distances, respectively. Indicates azimuth; Indicates the upper bound of the relative azimuth angle;

[0113] Based on the relative distance and relative azimuth, the tracking error between the actual trajectory and the desired trajectory is defined and obtained, and the expression for the tracking error is:

[0114] ,

[0115] In the formula: Indicates distance tracking error; Represents the expected trajectory distance and ; This indicates the azimuth tracking error. Indicates the desired azimuth angle and ;

[0116] S22: This embodiment designs a specified time performance function (PTFT) with an adaptive shape and a customizable convergence time. In marine missions, it is crucial to accurately control the ship to reach the desired trajectory within a specified time. The adaptive shape of the PTFT can flexibly adjust the system boundary according to the real-time state of the system and mission requirements, and customize the convergence time according to the target dynamics and search urgency to quickly lock onto the target.

[0117] The constructed time performance function is expressed as follows:

[0118]

[0119]

[0120]

[0121]

[0122] In the formula: A boundary performance function representing a specified time; express The first derivative; and Indicates a positive design parameter; , , Indicates the amount of intermediate parameters; express The first derivative; Indicates the set time; Indicates the quantity of intermediate parameters and ; and Indicates time design parameters; Indicates constant gain; Indicates the convergence time;

[0123] S23: Based on the constructed time performance function, obtain the constraints on the tracking error;

[0124] And the expression for the constraint condition is:

[0125]

[0126] In the formula: and Indicates the boundary gain coefficient;

[0127] S24: Obtain the error transformation function based on the constraints of the tracking error, wherein the expression of the error transformation function is...

[0128]

[0129]

[0130] In the formula: / Indicates intermediate parameter variables; , , , , , All represent design parameters; Represents the tracking error variable between the virtual ship and the following ship, and is The abbreviated form; Represents the error transformation function;

[0131] This embodiment also includes defining intermediate variables. and ,in express The estimated value;

[0132] Consider the following function:

[0133]

[0134] In the formula: and Represents the error transformation function; This represents the estimated value of the confounding variable; Representing the correlation coefficient, further obtained The derivative is used for subsequent simulation experiments for verification, and its expression is:

[0135]

[0136] In the formula: express The derivative; Representation and tracking error And the coefficient matrix or vector related to the system state variables are variables related to the control input, which are adjusted by the design parameters. Control capable of achieving convergence of the desired trajectory of the system state terms; express The derivative; express The derivative; This measures the distance between the unmanned vessel and the target. This indicates the error in estimating the unknown input gain.

[0137] S3: Obtain the system dynamics model including model uncertainties and unknown control input gains based on the unmanned ship mathematical model, and use the constructed unmanned ship data-driven neural predictor to estimate the model uncertainties and unknown control input gains to obtain the unmanned ship dynamics model;

[0138] This embodiment estimates system uncertainties, including unmodeled dynamics, unknown environmental disturbances, and unknown input gains, by developing a data-driven neural predictor (DDNP), thereby achieving dynamic and accurate estimation and tracking of system uncertainties.

[0139] Specifically, the following steps are included:

[0140] S31: Construct a data-driven neural predictor for unmanned ships based on the mathematical model of unmanned ships;

[0141] The unmanned vessel data-driven neural predictor includes a cascaded network unit composed of a neural network module and a filtering module, and a predictor development unit;

[0142] S311: Obtain the system dynamic equations of the mathematical model of the underactuated unmanned surface vessel, including model uncertainties. The expression of the system dynamic equations is as follows:

[0143]

[0144] In the formula: , , This represents the internal model uncertainty term of the unmanned vessel's mathematical model; , Indicates unknown control input gain; , Indicates system control input;

[0145] S312: Since the uncertainties of the unmanned vessel system model are difficult to obtain analytically, a neural network module is used to approximate the system dynamics equations. That is, a neural network is used to approximate the uncertainties of the unmanned vessel system model, and its expression is:

[0146]

[0147]

[0148] In the formula: Indicates the current time step; and Indicates the time step of the delay; Indicates transpose; This indicates the uncertainty of the system; Indicates period; Represents the Gaussian function; and Let represent the weight vector and approximation error of the neural network, respectively, and satisfy . and , and Indicates design constants;

[0149] S313: A filter is constructed using a filtering module to filter the uncertainties in the internal model after approximation processing. This is done to improve the stability of the neural network estimation. To extract the driving data features of the system and process the input data of the unmanned vessel system model, its expression is:

[0150]

[0151] ,

[0152]

[0153] In the formula: This represents an estimate of the gain for an unknown input. Indicates external input; This represents the filtered parameter vector; This represents the velocity derivative after filtering; Indicates a positive design constant; express The first derivative; Represents the filter parameter vector; express The initial value; express The filtered value and ; This represents a parameter used to compensate for system uncertainties. Indicates filter parameters; Represents the input feature vector of a neural network module ; The filter's parameter weight vector and ;

[0154] S314: In this embodiment, to enhance the learning process of the unmanned surface vessel (USV) data-driven neural predictor, a historical dataset of USV data-driven data stored in a preset memory stack is obtained, and the historical dataset is denoted as... ;

[0155] in, and , Indicates the stack length of the memory stack; Indicates time t=k The filter parameter vector at the location; Indicates time t=k The derivative of the filtering velocity at that point; Indicates time t=k Input gain estimate at the location;

[0156] S315: Based on cascaded network units, a data-driven neural predictor (DDNP) for unmanned surface vessels is constructed through a predictor development unit. Its expression is as follows:

[0157]

[0158]

[0159] In the formula: , , Represents design constants and ; express The estimated value and ; express The estimated value; Represents a positive definite coefficient matrix; This represents the adjustment factor for weight updates; This indicates the stack length of the memory stack; where memory stack storage is a data structure in computer science used to manage function calls and local variable storage during program execution; the stack is a Last In First Out (LIFO) data structure, meaning that the last data entered into the stack is the first to be retrieved. Memory stack storage plays a crucial role in program execution, especially in function calls, recursion, and local variable management; Represents the kinematic tracking error variable;

[0160] S316: An unmanned vessel data-driven neural predictor is constructed using reinforcement learning based on historical datasets. Its parameter weights are confirmed to reconstruct the unmanned vessel data-driven neural predictor, which is used to estimate the uncertainties of the model and obtain the dynamic model of the unmanned vessel. The method of reinforcement learning of the unmanned vessel data-driven neural predictor using historical datasets is a well-known existing technology and will not be described in detail here.

[0161] S4: Based on the error transformation function, construct a model-free data-driven virtual control law according to the dynamic model of the unmanned vessel;

[0162] Specifically, the expression for the constructed model-free data-driven virtual control law is as follows:

[0163]

[0164]

[0165]

[0166] In the formula: , , These represent virtual control laws for longitudinal, lateral, and steering directions, respectively. This represents an estimate of the unknown input gain. , Indicates the error transformation scaling factor; express The first derivative; , , , , , , , , Indicate design parameters; express The first derivative; Indicates intermediate parameters and , , ; , , Indicates the design scaling parameters; , , Represents the auxiliary variable used to adjust the dynamic behavior of errors in the control system, and the auxiliary system is... ; , This represents an adaptive parameter used to adjust the response strength of auxiliary system variables to control deviations and system states. Indicates control gain; Indicates intermediate parameters and , , ;

[0167] The model-free virtual control law constructed in this embodiment also includes letting The filtering process is performed using a second-order filter, and its expression is:

[0168]

[0169] in: These are filter parameters; they contain positive constants. and , making and ; express The derivative of .

[0170] Consider the following adaptive law:

[0171]

[0172] Combined energy function We can obtain:

[0173]

[0174] in , , ; , These are positive design parameters. Indicate design parameters and ;

[0175] S5: Define the kinematic tracking error variable based on the model-free data-driven virtual control law, and construct a data-driven controller based on the kinematic tracking error variable;

[0176] Specifically, the following steps are included:

[0177] S51: Based on the virtual control law and the unmanned vessel dynamics model, define the kinematic tracking error variable, whose expression is:

[0178]

[0179] In the formula: These represent the tracking error variables in the longitudinal, lateral, and steering directions, respectively.

[0180] S52: Construct a data-driven controller based on the kinematic tracking error variable; its expression is as follows:

[0181]

[0182]

[0183]

[0184]

[0185] In the formula: , These represent the longitudinal and steering direction control inputs, respectively. , The tracking error variables represent the longitudinal and steering directions; , Indicates intermediate variables; , This indicates the desired speed command that guides the system towards an ideal state. , Indicates control gain; , Describe the basis functions; , This represents an estimate of the unknown input gain. , Indicates the dynamic threshold parameter;

[0186] S6: Based on the data-driven controller design, a switching dynamic event triggering mechanism is used to switch between the fixed threshold triggering mechanism and the dynamic event triggering mechanism, and the unmanned vessel data-driven performance control at a specified time is realized based on the switching dynamic event triggering mechanism;

[0187] Specifically, the dynamic event triggering mechanism designed to switch between the fixed threshold triggering mechanism and the dynamic event triggering mechanism has the following expression:

[0188]

[0189]

[0190]

[0191] In the formula: Indicates in The data at any given moment drives the controller's output; Indicates the triggering error of switching dynamic event triggering mechanisms and ; and This indicates a dynamically adjustable threshold parameter; This represents the threshold parameter used to switch between the fixed threshold triggering mechanism FTTM and the dynamic event triggering mechanism DETM; , , Indicate design parameters; and Indicates a dynamically adjustable threshold parameter , The first derivative; Indicates time interval The actual control input within; Indicates time parameters;

[0192] This embodiment designs a data-driven controller and a model-free data-driven virtual control law based on a data-driven neural predictor. It also introduces a switching dynamic event triggering mechanism to accurately determine the control input of the unmanned vessel system. When the error exceeds the tolerance range or the system state changes abruptly, the control input is quickly adjusted. When the system is stable within the accuracy range, the control signal is kept stable to avoid over-adjustment and oscillation, thus achieving precise control of the unmanned vessel.

[0193] This embodiment also includes a stability analysis process:

[0194] System stability analysis is performed by constructing a Lyapunov function, and relevant constants are then determined to ensure stable system operation. Specifically, this involves constructing the Lyapunov function. ;

[0195] in, , , ;

[0196] Through the and By taking the derivative, we can eventually obtain The relationship between time t and the change of time t:

[0197]

[0198] In the formula: , ,

[0199] , , ; Indicates the control gain parameter; Indicates the error feedback gain; This represents the total amount of the upper bound of Lyapunov; This represents the minimum bound for the Lyapunov convergence rate; Indicates the range of influence of the error; The basis functions of a neural network; Indicates the first k The estimation error of a historical sample; Indicate design parameters; This represents the parameters related to the estimation of unknown input gain; Indicates state-related parameters; These represent the system state estimation parameters. It can be inferred that as time approaches infinity, tending to By selecting appropriate design parameters, it is possible to The error is arbitrarily small; therefore, the error of the closed-loop system is SGUUB. In this embodiment, to prevent... The behavior and sampling interval must meet the following requirements. Furthermore, we can obtain:

[0200]

[0201] in: express The upper bound, and >0; Indicates the sampling interval in the event triggering mechanism; This represents a fixed threshold parameter in the triggering mechanism; therefore, the minimum interval between events is positive, excluding... Phenomenon.

[0202] The beneficial effects of the method described in this embodiment are as follows:

[0203] 1. By constructing an unmanned vessel data-driven neural predictor, the model uncertainty and position input gain in the system dynamics model are estimated, thus obtaining the unmanned vessel dynamics model. Based on this, a model-free data-driven virtual control law is designed. This unmanned vessel data-driven neural predictor can adapt to new tasks, model changes, and environmental changes without relying on prior information of model parameters. Unlike traditional deterministic learning methods that are limited to repetitive tasks, this method is more flexible and adaptable.

[0204] 2. The constructed time-defined performance function has shape adaptability and preset time convergence. Its convergence time can be defined by the user and is independent of the system initial conditions and system parameters. Compared with previous work, this time-defined performance function can be applied to unmanned ships without fixed data models, enhancing its adaptability to different environments.

[0205] 3. The designed switching dynamic event triggering mechanism can effectively reduce the communication burden. By dynamically switching between the fixed threshold triggering mechanism and the dynamic event triggering mechanism, communication resources are further saved. The advantages are more obvious, especially when the actuator jumps or changes suddenly during the operation of the controller.

[0206] The control architecture of this embodiment for a model-free, data-driven, time-based performance control method for unmanned surface vessels is as follows: Figure 2 As shown, the architecture encompasses multiple components, including kinematic loop-defined performance control, auxiliary systems, leader tracking error constraints, error transformation, neural update rate, data-driven DDNP for the unmanned surface vessel (USV), external disturbance filters, memory stacks, and the fixed threshold triggering mechanism and dynamic event triggering mechanism of SDETM. These components work together to achieve effective control of the underactuated USV. Simulation experiments were conducted to verify the effectiveness of the method. A series of parameters were set, including desired trajectory parameters, controller design parameters, external disturbances, and initial conditions. Figure 3 The planar trajectory is shown, and the trajectory tracking control plane of the method in this embodiment is compared with that of the existing algorithm (the algorithm without data driving). It is clear that the method in this embodiment can track the desired trajectory more accurately, which demonstrates the superiority of the algorithm in actual path planning. Figure 4 and Figure 5 The positional and angular errors are presented separately. Compared with existing algorithms, the method in this embodiment has more stable errors, can better meet the control accuracy requirements, and ensure the accuracy of the unmanned vessel during operation. Figure 6 The approximation accuracy of DDNP is demonstrated, reflecting its effectiveness in estimating system uncertainties and providing a reliable basis for the design of adaptive control laws. Figure 7 The estimation results of the unknown input gain are given, which reflects the method's ability to learn and master the unknown parameters of the system. Figure 8 Displaying the actual control input shows the changes in the control signal, which helps in analyzing the effectiveness of the control strategy. Figure 9 Presenting the time interval under SDETM demonstrates the mechanism's effectiveness in reducing communication resource consumption and ensuring stable system operation while efficiently utilizing resources.

[0207] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A model-free data-driven fixed-time performance control method for unmanned ships, characterized in that, Specifically, the following steps are included: S1: Obtain the underactuated unmanned vessel mathematical model containing model uncertainties; S2: Define and obtain the tracking error between the actual trajectory and the expected trajectory based on the mathematical model of the unmanned vessel; And obtain the error transformation function based on the constructed time performance function; S3: Obtain the system dynamics model including model uncertainties and unknown control input gains based on the unmanned ship mathematical model, and use the constructed unmanned ship data-driven neural predictor to estimate the model uncertainties and unknown control input gains to obtain the unmanned ship dynamics model; The unmanned vessel data-driven neural predictor includes a cascaded network unit composed of a neural network module and a filtering module, and a predictor development unit; S4: Based on the error transformation function, construct a model-free data-driven virtual control law according to the dynamic model of the unmanned vessel; S5: Define the kinematic tracking error variable based on the model-free data-driven virtual control law, and construct a data-driven controller based on the kinematic tracking error variable; S6: Based on the data-driven controller design, a switching dynamic event triggering mechanism is used to switch between the fixed threshold triggering mechanism and the dynamic event triggering mechanism, and the unmanned vessel data-driven performance control at a specified time is realized based on the switching dynamic event triggering mechanism; The designed dynamic event triggering mechanism for switching between fixed threshold triggering and dynamic event triggering is expressed as follows: In the formula: Indicates in The data at any given moment drives the controller's output; Indicates the triggering error of switching dynamic event triggering mechanisms and ; and This indicates a dynamically adjustable threshold parameter; This represents the threshold parameter used to switch between the fixed threshold triggering mechanism FTTM and the dynamic event triggering mechanism DETM; , , Indicate design parameters; and Indicates a dynamically adjustable threshold parameter , The first derivative; Indicates time interval The actual control input within; Indicates the time parameter.

2. The time-based performance control method for model-free unmanned surface vessels (USVs) based on data-driven methods according to claim 1, characterized in that, The expression for the underactuated unmanned surface vessel mathematical model, including model uncertainties, obtained in S1 is as follows: In the formula: This indicates the unmanned vessel's x-coordinate, y-coordinate, and bow angle in a preset Earth coordinate system. This indicates the longitudinal velocity, lateral velocity, and angular velocity of the unmanned surface vessel. Indicates the internal model uncertainty term; This indicates unknown disturbances in the environment and , Represents the upper bound of unknown interference; This indicates the control inputs provided by the thrusters and rudders; express The first derivative; express The first derivative; , This indicates the added quality of the model.

3. The time-based performance control method for model-free unmanned surface vessels (USVs) based on data-driven methods according to claim 2, characterized in that, S2 specifically includes the following steps: S21: Obtain the relative distance and relative azimuth angle based on the unmanned surface vessel's mathematical model; its expression is as follows: In the formula: Indicates relative distance and , ; These represent the minimum and maximum preset relative distances, respectively. Indicates azimuth; Indicates relative azimuth and and ; Indicates the upper bound of the relative azimuth angle; Based on the relative distance and relative azimuth, the tracking error between the actual trajectory and the desired trajectory is defined and obtained, and the expression for the tracking error is: , In the formula: Indicates distance tracking error; Represents the expected trajectory distance and ; This indicates the azimuth tracking error. Indicates the desired azimuth angle and ; S22: Construct the specified time performance function, the expression of which is: In the formula: A boundary performance function representing a specified time; express The first derivative; and Indicates a positive design parameter; , , Indicates the amount of intermediate parameters; express The first derivative; Indicates the set time; Indicates the quantity of intermediate parameters and ; and Indicates time design parameters; Indicates constant gain; Indicates the convergence time; S23: Based on the constructed time performance function, obtain the constraints on the tracking error; And the expression for the constraint condition is: In the formula: and Indicates the boundary gain coefficient; S24: Obtain the error transformation function based on the constraints of the tracking error, wherein the expression of the error transformation function is... In the formula: / Indicates intermediate parameter variables; , , , , , All represent design parameters; Represents the tracking error variable between the virtual ship and the following ship, and is The abbreviated form; This represents the error transformation function.

4. The time-based performance control method for model-free unmanned surface vessels (USVs) based on data-driven methods according to claim 3, characterized in that, S3 specifically includes the following steps: S31: Construct a data-driven neural predictor for unmanned ships based on the mathematical model of unmanned ships; S311: Obtain the system dynamic equations of the mathematical model of the underactuated unmanned surface vessel, including model uncertainties. The expression of the system dynamic equations is as follows: In the formula: , , This represents the internal model uncertainty term of the unmanned vessel's mathematical model; , Indicates unknown control input gain; , Indicates system control input; S312: The system dynamics equations are approximated using a neural network module, and the expression is: In the formula: Indicates the current time step; and Indicates the time step of the delay; Indicates transpose; This indicates the uncertainty of the system; Indicates period; Represents the Gaussian function; and Let represent the weight vector and approximation error of the neural network, respectively, and satisfy . and , and Indicates design constants; S313: A filter is constructed using the filtering module to filter the uncertainties in the internal model after approximation processing. Its expression is: , In the formula: This represents an estimate of the gain for an unknown input. Indicates external input; This represents the filtered parameter vector; This represents the velocity derivative after filtering; Indicates a positive design constant; express The first derivative; Represents the filter parameter vector; express The initial value; express The filtered value and ; This represents a parameter used to compensate for system uncertainties. Indicates filter parameters; Represents the input feature vector of a neural network module ; The filter's parameter weight vector and ; S314: Obtain the historical dataset driven by unmanned vessel data stored in the preset memory stack; And the historical dataset is denoted as ; in, and , Indicates the stack length of the memory stack; Indicates time t=k The filter parameter vector at the location; Indicates time t=k The derivative of the filtering velocity at that point; Indicates time t =k Input gain estimate at the location; S315: Based on cascaded network units, a data-driven neural predictor for unmanned surface vessels is constructed through a predictor development unit. Its expression is as follows: In the formula: , , Represents design constants and ; express The estimated value and ; express The estimated value; Represents a positive definite coefficient matrix; This represents the adjustment factor for weight updates; Indicates the stack length of the memory stack; Represents the kinematic tracking error variable; S316: An unmanned vessel data-driven neural predictor is constructed using reinforcement learning based on historical datasets. Its parameter weights are confirmed to reconstruct the unmanned vessel data-driven neural predictor, which is used to estimate the uncertainties of the model and obtain the dynamic model of the unmanned vessel.

5. A time-based performance control method for model-free unmanned surface vessels (USVs) based on data-driven methods according to claim 4, characterized in that, The expression for the model-free data-driven virtual control law constructed in S4 is: In the formula: , , These represent virtual control laws for longitudinal, lateral, and steering directions, respectively. This represents an estimate of the unknown input gain. , Indicates the error transformation scaling factor; express The first derivative; , , , , , , , , Indicate design parameters; express The first derivative; Indicates intermediate parameters and , , ; , , Indicates the design scaling parameters; , , Represents the auxiliary variable used to adjust the dynamic behavior of errors in the control system, and the auxiliary system is... ; , This represents an adaptive parameter used to adjust the response strength of auxiliary system variables to control deviations and system states. Indicates control gain; Indicates intermediate parameters and , , .

6. A time-based performance control method for model-free unmanned surface vessels (USVs) based on data-driven methods according to claim 5, characterized in that, S5 specifically includes the following steps: S51: Based on the virtual control law and the unmanned vessel dynamics model, define the kinematic tracking error variable, whose expression is: In the formula: These represent the tracking error variables in the longitudinal, lateral, and steering directions, respectively. S52: Construct a data-driven controller based on the kinematic tracking error variable; its expression is as follows: In the formula: , These represent the longitudinal and steering direction control inputs, respectively. , The tracking error variables represent the longitudinal and steering directions; , Indicates intermediate variables; , This indicates the desired speed command that guides the system towards an ideal state. , Indicates control gain; , Describe the basis functions; , This represents an estimate of the unknown input gain. , This represents the dynamic threshold parameter.

Images