Target object control method, device, computer equipment and storage medium

By using the target neural network to estimate and compensate prediction of nonlinear terms in the target object control method, the control accuracy problem caused by ocean perturbation in traditional technology is solved, and more accurate target object control is achieved.

CN117666341BActive Publication Date: 2024-11-12TIANHE TRAILBLAZER PHOTOVOLTAIC STENT (JIANGSU CHANGZHOU) CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202311622392.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-11-30
Publication Date
2024-11-12
Estimated Expiration
2043-11-30

AI Technical Summary

Technical Problem

In traditional technology, the control of the target object in the marine environment is difficult to accurately determine, because marine disturbances in the external environment cause unknown interference, making it difficult to obtain accurate mathematical model parameters, affecting the control accuracy.

Method used

The target object control method is adopted, including determining the integral sliding mode terms and nonlinear terms in the driving control law, using the target neural network to iteratively perform the estimation compensation prediction of nonlinear terms based on speed tracking error, and adjusting the neural network structure to generate more accurate control Signal.

Benefits of technology

It improves the control accuracy of the target object, can drive the target object to a preset position more accurately, and enhances the stability and robustness of control in the marine environment.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117666341B_ABST
    Figure CN117666341B_ABST
Patent Text Reader

Abstract

The present application relates to a target object control method, device, computer equipment and storage medium. The method includes: determining a driving control law for a target object; determining a speed tracking error based on a first position and a first speed of the target object at a first moment, and an expected position corresponding to the target object at a second moment; determining the value of an integral sliding mode term based on the speed tracking error; using a target neural network to iteratively estimate, compensate and predict a nonlinear term based on the speed tracking error, and in the process of estimating, compensating and predicting, iteratively adjusting the number of neurons in the target neural network based on the activation corresponding to the target neural network until the iteration stops, and obtaining an estimated value corresponding to the nonlinear term; substituting the value and estimated value of the integral sliding mode term into the driving control law to generate a control signal; the control signal is used to control the target object to travel to the expected position at the second moment. The use of this method can improve the control accuracy of the target object.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of automatic control technology, and in particular to a target object control method, device, computer equipment and storage medium. Background Art

[0002] With the development of science and technology, in order to further explore the natural environment, automatic control technology for target objects has emerged. The core of this technology is to control the target object to travel to the corresponding designated position at each designated time, so that the actual driving trajectory of the target object conforms to the preset trajectory. For example, in the marine environment, the target object can be an underactuated surface ship, and the exploration mission can be performed by controlling the underactuated surface ship.

[0003] In traditional technology, a mathematical model is usually established for the target object, and the target object is controlled based on the control signal generated by the mathematical model. However, in the actual driving process, there are usually unknown interferences in the external environment, such as ocean disturbances in the marine environment. In the presence of these interferences, it is difficult to obtain accurate mathematical model parameters, resulting in inaccurate control of the target object. Summary of the invention

[0004] Based on this, it is necessary to provide a target object control method, device, computer equipment, computer-readable storage medium and computer program product that can improve control accuracy in response to the above technical problems.

[0005] In the first aspect, the present application provides a target object control method. The method includes: determining a driving control law for the target object; the driving control law includes an integral sliding mode term and a nonlinear term; determining a speed tracking error based on the first position and the first speed of the target object at the first moment, and the expected position of the target object at the second moment; the second moment is the next moment of the first moment; determining the value of the integral sliding mode term based on the speed tracking error; using a target neural network to iteratively estimate, compensate and predict the nonlinear term based on the speed tracking error, and in the process of estimating, compensating and predicting, iteratively adjusting the number of neurons in the target neural network based on the activation corresponding to the target neural network until the iteration stops, and obtaining an estimated value corresponding to the nonlinear term; substituting the value of the integral sliding mode term and the estimated value into the driving control law to generate a control signal; the control signal is used to control the target object to travel to the expected position at the second moment.

[0006] In the second aspect, the present application also provides a target object control device. The device includes: a control law determination module, which is used to determine the driving control law for the target object; the driving control law includes an integral sliding mode term and a nonlinear term; a speed error determination module, which is used to determine the speed tracking error based on the first position and the first speed of the target object at the first moment, and the expected position of the target object at the second moment; the second moment is the next moment of the first moment; a sliding mode calculation module, which is used to determine the value of the integral sliding mode term based on the speed tracking error; an estimation compensation module, which is used to use the target neural network to iteratively estimate, compensate and predict the nonlinear term based on the speed tracking error, and in the estimation compensation prediction process, iteratively adjust the number of neurons in the target neural network based on the activation degree corresponding to the target neural network until the iteration stops, and obtain the estimated value corresponding to the nonlinear term; a control signal generation module, which is used to substitute the value of the integral sliding mode term and the estimated value into the driving control law to generate a control signal; the control signal is used to control the target object to travel to the expected position at the second moment.

[0007] In some embodiments, the estimation and compensation module is also used to: determine the activation degree corresponding to each of the neurons based on the activation function of the target neural network and the speed tracking error during each round of iteration of the estimation and compensation prediction; adjust the number of neurons in the target neural network based on the activation degree of each neuron to obtain an adjusted target neural network; determine the weight parameters corresponding to the adjusted target neural network based on the neuron weights corresponding to each neuron in the adjusted target neural network; stop iteration when the weight parameters reach the ideal weight parameters, and use the target neural network corresponding to the ideal weight parameters to determine the estimated value corresponding to the nonlinear term based on the speed tracking error.

[0008] In some embodiments, the estimation compensation module is also used to: determine the target activation from the activations corresponding to each of the neurons; when the target activation is less than or equal to a first threshold, add new neurons to the target neural network to obtain an adjusted target neural network; wherein the neuron weights corresponding to the new neurons are determined based on the neuron weights corresponding to the neurons to which the target activation belongs.

[0009] In some embodiments, the estimation compensation module is also used to: for each neuron, calculate the index value corresponding to the neuron based on the activation degree corresponding to the neuron; when the index value of the neuron is less than or equal to a second threshold, delete the neuron to obtain an adjusted target neural network.

[0010] In some embodiments, the speed error determination module is also used to: determine a virtual speed control law for the target object; substitute the first position and the expected position into the virtual speed control law to obtain the expected speed corresponding to the target object; and determine the speed tracking error based on the difference between the first speed and the expected speed.

[0011] In some embodiments, the control law determination module is also used to: establish a dynamic model with input time lag of the control signal for the target object; perform model conversion on the dynamic model to obtain a differential equation corresponding to the speed tracking error; the differential equation includes the input time lag; determine an integral sliding surface for the speed tracking error; and obtain a driving control law with the input time lag based on the integral sliding surface and the differential equation corresponding to the speed tracking error.

[0012] In some embodiments, the driving control law includes a surge control law and a yaw control law; the speed tracking error includes a surge speed tracking error and a yaw speed tracking error; the integral sliding surface includes a first-order integral sliding surface corresponding to the surge speed tracking error and a second-order integral sliding surface corresponding to the yaw speed tracking error; the control law determination module is further used to: obtain the surge control law with the input time lag based on the first-order integral sliding surface and the differential equation corresponding to the surge speed tracking error; obtain the yaw control law with the input time lag based on the second-order integral sliding surface and the differential equation corresponding to the yaw speed tracking error.

[0013] In some embodiments, the target object is an under-actuated surface vessel; after generating the control signal, the device is further used to: send the control signal to the under-actuated surface vessel; the under-actuated surface vessel receives the control signal and travels to the desired position at the second moment according to the control signal.

[0014] In a third aspect, the present application further provides a computer device, wherein the computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps in the target object control method when executing the computer program.

[0015] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps in the target object control method are implemented.

[0016] In a fifth aspect, the present application further provides a computer program product, wherein the computer program product comprises a computer program, and when the computer program is executed by a processor, the steps in the target object control method are implemented.

[0017] The above-mentioned target object control method, device, computer equipment, storage medium and computer program product calculate complex nonlinear terms in the driving control law, use the target neural network to iteratively estimate, compensate and predict the nonlinear terms based on the speed tracking error. During the estimation, compensation and prediction process, the number of neurons in the target neural network is iteratively adjusted based on the activation degree corresponding to the target neural network until the iteration stops, and the estimated value corresponding to the nonlinear term is obtained. Since the number of neurons in the target neural network can be updated online in real time, the estimated compensation result is more accurate, and the value and estimated value of the integral sliding mode term are substituted into the driving control law, and the generated control signal is more accurate, so that the control signal can be used to more accurately control the target object. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 An application environment diagram of a target object control method in an embodiment;

[0019] Figure 2 A schematic diagram of a flow chart of a target object control method in one embodiment;

[0020] Figure 3 A schematic diagram of a process of using a target neural network to perform estimation compensation prediction in one embodiment;

[0021] Figure 4 A schematic diagram of trajectory tracking of an underactuated surface vessel in one embodiment;

[0022] Figure 5 A control block diagram of an underactuated surface vessel in one embodiment;

[0023] Figure 6 A simulation schematic diagram of a control system in one embodiment;

[0024] Figure 7 A schematic diagram of a change in position tracking error with input time lag in one embodiment;

[0025] Figure 8 A schematic diagram of a change in a speed tracking error with an input time lag in one embodiment;

[0026] Fig. 9 A schematic diagram of control force and torque with input lag in one embodiment;

[0027] Fig.10 It is a schematic diagram of an estimation curve of a first nonlinear term of a target neural network in one embodiment;

[0028] Fig.11 It is a schematic diagram of an estimation curve of the second nonlinear term of the target neural network in one embodiment;

[0029] Fig.12 is a structural block diagram of a target object control device in one embodiment;

[0030] Fig.13 is an internal structure diagram of a computer device in one embodiment;

[0031] Fig.14 FIG. 4 is a diagram showing the internal structure of a computer device in another embodiment. DETAILED DESCRIPTION

[0032] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0033] The target object control method provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown, the application environment includes a controller 102, a target object 104, and a computer device 106. The controller 102 is used to generate a control signal for the target object 104. The computer device 106 can store the control signal generated by the controller 102 and can also provide a driving control law for the controller 102. The controller 102 can communicate with the target object 104 and the computer device 106, for example, through a network.

[0034] The computer device 106 may be a terminal, including but not limited to various personal computers, laptops, smart phones, and tablet computers. The computer device 106 may also be a server, which may be implemented by an independent server or a server cluster consisting of multiple servers.

[0035] Those skilled in the art will understand that Figure 1 The application environment shown in is only a part of the scenarios related to the present application solution and does not constitute a limitation on the application environment of the present application solution.

[0036] In some embodiments, Figure 2 As shown, a target object control method is provided, which is applied to Figure 1 The controller 102 in the example is used as an example to illustrate, and the following steps are included:

[0037] Step 202, determining a driving control law for the target object; the driving control law includes an integral sliding mode term and a nonlinear term.

[0038] The target object is an object that is controlled to travel along a preset trajectory. For example, the target object can be an unmanned vehicle or an underactuated surface ship. Therefore, the target object control method can also be called a trajectory tracking control method for the target object. The driving control law refers to an algorithm used to form a control signal for the target object, which describes the functional relationship between the controlled state variable and the control signal. The integral sliding mode term is determined using the integral sliding mode control theory, and the nonlinear term characterizes the external disturbances that exist during the driving process of the target object.

[0039] Specifically, the controller may obtain the driving control law for the target object from a computer device. The computer device pre-stores the formal control law of the target object, and the computer device may be a terminal or a server.

[0040] Step 204, determining a velocity tracking error based on a first position and a first velocity of the target object at a first moment, and an expected position of the target object corresponding to a second moment; the second moment is a moment next to the first moment.

[0041] Among them, the first moment is any moment in the driving process of the target object, which can be understood as the current moment, and the second moment is the next moment of the first moment. The first position refers to the real-time position of the target object at the first moment, which can be the coordinate in the ground coordinate system. The first speed refers to the real-time speed of the target object at the first moment. The expected position refers to the position that the target object needs to reach at the second moment, which can be determined from the preset driving trajectory using the second moment. The preset driving trajectory is predetermined and is a preset path from the starting position to the target position, consisting of multiple specified positions corresponding to specified moments. The speed tracking error characterizes the difference between the first speed and the ideal expected speed.

[0042] Specifically, the controller obtains the first position and the first speed corresponding to the target object at the first moment, and obtains the expected position corresponding to the target object at the second moment from the preset driving trajectory. The controller can also obtain a virtual speed control law for the target object, and then substitute the first position and the expected position into the virtual speed control law to obtain the expected speed corresponding to the target object, and determine the speed tracking error according to the difference between the first speed and the expected speed. Among them, the virtual speed control law is an algorithm for calculating the ideal expected speed, which is predetermined. The expected speed is the speed that the target object needs to reach the expected position at the second moment.

[0043] Step 206, determining the value of the integral sliding mode term based on the speed tracking error;

[0044] The integral sliding mode term is an integral sliding mode surface designed based on the differential equation of the speed tracking error of the target object using the sliding mode control method, and is used to solve the convergence problem of the speed tracking error.

[0045] Specifically, the controller obtains an expression of the integral sliding mode term, substitutes the speed tracking error into the expression of the integral sliding mode term, and obtains the value of the integral sliding mode term.

[0046] Step 208, use the target neural network to iteratively estimate, compensate and predict the nonlinear term based on the speed tracking error. During the estimation, compensation and prediction process, the number of neurons in the target neural network is iteratively adjusted based on the activation corresponding to the target neural network until the iteration stops, and the estimated value corresponding to the nonlinear term is obtained.

[0047] The target neural network is used to approximate nonlinear functions, so the nonlinear terms can be estimated, compensated and predicted. The target neural network includes multiple neurons. Compared with traditional neural networks, the target neural network can update the number of neurons online, change the network structure, and achieve better function approximation performance.

[0048] Specifically, the controller can input the speed tracking error into the target neural network for estimation compensation prediction, and output the estimated value corresponding to the nonlinear term. The estimation compensation prediction is an iterative process. In each round of iteration, the activation degree corresponding to each neuron is determined, and the number of neurons is adjusted based on the activation degree corresponding to each neuron. The iteration is stopped when the weight parameters determined based on each neuron in the adjusted target neural network reach the ideal weight parameters, and the target neural network corresponding to the ideal weight parameters is used to determine the estimated value corresponding to the nonlinear term based on the speed tracking error.

[0049] Step 210, substitute the value and estimated value of the integral sliding mode term into the driving control law to generate a control signal; the control signal is used to control the target object to travel to the desired position at the second moment.

[0050] The control signal is usually a control voltage signal.

[0051] Specifically, the controller substitutes the value of the integral sliding mode term and the estimated value corresponding to the nonlinear term into the driving control law to obtain a control signal that causes the target object to travel to the desired position at the second moment. The controller can send the control signal to the actuator of the target object. The actuator of the target object receives the control signal and executes according to the control signal to generate at least one of a propulsion force or a steering force that causes the target object to travel to the desired position.

[0052] In the above-mentioned target object control method, by calculating the complex nonlinear terms in the driving control law, the target neural network is used to iteratively estimate, compensate and predict the nonlinear terms based on the speed tracking error. During the estimation, compensation and prediction process, the number of neurons in the target neural network is iteratively adjusted based on the activation corresponding to the target neural network until the iteration stops, and the estimated value corresponding to the nonlinear term is obtained. Since the number of neurons in the target neural network can be updated online in real time, the estimated compensation result is more accurate, and the value and estimated value of the integral sliding mode term are substituted into the driving control law, and the generated control signal is more accurate, so that the control signal can be used to more accurately control the target object.

[0053] In some embodiments, during the estimation compensation prediction process, the number of neurons in the target neural network is iteratively adjusted based on the activation corresponding to the target neural network until the iteration stops to obtain an estimated value corresponding to the nonlinear term, including: during each round of iteration of the estimation compensation prediction, the activation corresponding to each neuron is determined based on the activation function of the target neural network and the speed tracking error; the number of neurons in the target neural network is adjusted based on the activation of each neuron to obtain an adjusted target neural network; the weight parameters corresponding to the adjusted target neural network are determined based on the neuron weights corresponding to each neuron in the adjusted target neural network; the iteration is stopped when the weight parameter reaches the ideal weight parameter, and the target neural network corresponding to the ideal weight parameter is used to determine the estimated value corresponding to the nonlinear term based on the speed tracking error.

[0054] The activation function is used to calculate the output of the neuron, i.e., the activation degree, which is used to characterize the degree of activation of the neuron. The weight parameter is used to calculate the output of the target neural network, i.e., the estimated value corresponding to the nonlinear term. The ideal weight parameter refers to the value of the weight parameter when it reaches the ideal state.

[0055] Specifically, the controller inputs the speed tracking error into the target neural network for estimation, compensation and prediction. In each round of iteration of the target neural network for estimation, compensation and prediction, for each neuron, the speed tracking error and the neuron parameter corresponding to the neuron are substituted into the activation function to obtain the activation degree corresponding to each neuron. Then, based on the activation degree corresponding to each neuron, it is determined whether to add new neurons or delete neurons, thereby adjusting the number of neurons in the target neural network to obtain the adjusted target neural network.

[0056] In some embodiments, a target neural network is used to perform function approximation on a nonlinear function, which can be expressed as:

[0057] f(x)=W *T N(x)+ε(x)

[0058] Where x is the input vector. *is the ideal weight parameter of the target neural network. N(x)=[N1(x), N2(x),…, N n (x)] T , represents the neuron vector composed of the activations of each neuron. ε(x) represents the approximation error of the target neural network, satisfying ||ε(x)||≤ε m , where ε m is a small normal number.

[0059] In some embodiments, the ideal weight parameter W * It can be expressed as:

[0060]

[0061] in, is an estimate of W, is the estimated error of W, satisfying ||W||≤W * .

[0062] In some embodiments, a Gaussian function may be used as the activation function of the target neural network. Assume that the formula for calculating the activation degree using the activation function is as follows:

[0063]

[0064] Where j is the jth neuron node in the hidden layer of the target neural network, b j represents the width of the target neural network, c j is the center point vector of the jth neuron node.

[0065] In this embodiment, the number of neurons in the target neural network is adjusted based on the activation degree of each neuron in the estimation compensation prediction process to obtain an adjusted target neural network, thereby optimizing the structure of the neural network online and obtaining ideal weight parameters, thereby using the target neural network corresponding to the ideal weight parameters to determine a more accurate estimate based on the speed tracking error.

[0066] In some embodiments, the number of neurons in the target neural network is adjusted based on the activation of each neuron to obtain an adjusted target neural network, including: determining a target activation from the activations corresponding to each neuron; when the target activation is less than or equal to a first threshold, adding a new neuron to the target neural network to obtain an adjusted target neural network; wherein the neuron weight corresponding to the new neuron is determined based on the neuron weight corresponding to the neuron to which the target activation belongs.

[0067] The target activation is determined from the activations of the neurons, and represents the activation level of the target neural network. The first threshold is preset to a value between 0 and 1, and is used to determine whether the activation level of the target neural network reaches the ideal level.

[0068] Specifically, the controller can determine the target activation degree from the activation degrees corresponding to each neuron. For example, the maximum activation degree among the activation degrees corresponding to each neuron can be used as the target activation degree, and the target activation degree can also be called the maximum degree. The controller then compares the target activation degree with the first threshold. When the target activation degree is greater than the first threshold, it is determined that the activation degree of the target neural network has reached the ideal activation degree. When the target activation degree is less than the first threshold, it is determined that the activation degree of the target neural network has not reached the ideal activation degree, and new neurons are added to the target neural network to obtain an adjusted target neural network. For example, the activation degree of a neuron is expressed as N k , V th is the first threshold, V th ∈(0,1), then the target activation N m It is expressed as:

[0069]

[0070] Where h is the number of neurons in the target neural network; if N m ≤V th , new neurons need to be added.

[0071] In some embodiments, the neuron weight corresponding to the new neuron can be determined based on the neuron weight corresponding to the neuron to which the target activation belongs. For example, the new neuron is represented by h new , the neuron weights corresponding to the new neurons are defined as follows:

[0072]

[0073] in, is the center point vector corresponding to the new neuron. m ,c m ,b m is the parameter in the activation function corresponding to the neuron to which the target activation belongs. is the neuron weight corresponding to the new neuron. is the initial index value corresponding to the new neuron.

[0074] In this embodiment, by adding new neurons to the target neural network when the target activation degree is less than or equal to the first threshold, an adjusted target neural network is obtained, so that the activation degree corresponding to the target neural network can maintain the ideal activation degree, thereby improving the computing performance of the target neural network and thus improving the computing accuracy.

[0075] In some embodiments, the number of neurons in the target neural network is adjusted based on the activation level of each neuron to obtain an adjusted target neural network, including: for each neuron, calculating the index value corresponding to the neuron based on the activation level corresponding to the neuron; when the index value of the neuron is less than or equal to a second threshold, deleting the neuron to obtain the adjusted target neural network.

[0076] The index value is continuously updated based on the activation degree corresponding to the neuron, and the initial value of the index value is 1. The second threshold is a pre-set positive constant used to determine whether the neuron should be deleted.

[0077] Specifically, for each neuron, the controller can use a preset index value algorithm to calculate the index value corresponding to the neuron based on the activation degree corresponding to the neuron, and then compare the index value with the second threshold value. If the index value is greater than the second threshold value, the neuron is retained; if the index value is less than or equal to the second threshold value, the neuron is deleted until all neurons are traversed to obtain the adjusted target neural network. For example, assuming I th is the preset threshold, I r is the index value of the rth neuron, I th is the second threshold, if I r ≤I th , then delete the rth neuron.

[0078] In some embodiments, the preset index value algorithm can be expressed by the following formula:

[0079]

[0080] in, is the index value of the current rth neuron. r is the activation degree of the rth neuron. th and ι are positive constants.

[0081] In this embodiment, the index value corresponding to the neuron is calculated based on the activation degree corresponding to the neuron. When the index value of the neuron is less than or equal to the second threshold, the neuron is deleted to obtain an adjusted target neural network. When the network structure is too complex, too many neurons are deleted, thereby streamlining the network structure and improving computing efficiency.

[0082] In some embodiments, Figure 3As shown, a flow chart of using the target neural network for estimation compensation prediction is shown. First, the controller calculates the virtual speed and speed tracking error, takes the speed tracking error as the input of the target neural network, and then obtains the output of each neuron in the target neural network, that is, the activation degree, and determines the maximum degree from the activation degree, and then determines whether the maximum degree is greater than the first threshold, so as to determine whether to add neurons, and then calculates the index value corresponding to each neuron, and determines whether the index value of each neuron is greater than the second threshold, so as to determine whether to delete the neuron, until the control ends and the iterative output stops.

[0083] In some embodiments, a speed tracking error is determined based on a first position and a first speed of the target object at a first moment, and an expected position corresponding to the target object at a second moment, including: determining a virtual speed control law for the target object; substituting the first position and the expected position into the virtual speed control law to obtain the expected speed corresponding to the target object; and determining the speed tracking error according to the difference between the first speed and the expected speed.

[0084] Among them, the virtual velocity control law is used to perform stability control on the error in the kinematic model of the target object.

[0085] In some embodiments, the step of determining the virtual speed control law includes: defining a position tracking error representing the error between the actual position of the target object and the expected position, and determining a differential equation corresponding to the position tracking error, thereby designing a virtual speed control law that enables the position tracking error to converge within a specified time based on the differential equation corresponding to the position tracking error. For example, x represents the horizontal coordinate of the target object, y represents the vertical coordinate of the target object, and the position vector representing the expected position is defined as η d =[x d ,y d ] T , the position tracking error is:

[0086]

[0087] Among them, x e and e is the position tracking error.

[0088] Deriving the above equation, the differential equation corresponding to the position tracking error is constructed as follows:

[0089]

[0090] According to the differential equation corresponding to the position tracking error, the designed virtual speed control law is expressed by the following formula:

[0091]

[0092] Among them, u and υ are the true longitudinal and lateral velocities of the target object, respectively, and ψ is the angular velocity of the target object. α u and α υ They represent the expected surge velocity and the expected lateral surge velocity respectively. is the derivative of the position vector of the desired position. x , o x , k1 and l y , o y , k2 are the constant coefficients in the function.

[0093] In this embodiment, by determining a virtual speed control law that can make the position tracking error of the target object converge within a specified time, the desired speed is accurately calculated based on the virtual speed control law, thereby obtaining the accurate speed tracking error.

[0094] In some embodiments, determining a driving control law for a target object includes: establishing a dynamic model with an input time lag of a control signal for the target object; performing model conversion on the dynamic model to obtain a differential equation corresponding to a speed tracking error; including the input time lag in the differential equation; determining an integral sliding surface for the speed tracking error; and obtaining a driving control law with an input time lag based on the integral sliding surface and the differential equation corresponding to the speed tracking error.

[0095] Among them, input lag refers to the delay time from the controller generating a control signal to the target object moving according to the control signal, which is usually caused by the actuator of the target object.

[0096] Specifically, when the actuator receives the control signal and executes it, there is a certain input lag, and the input lag will cause the position and posture of the target object to change laggingly. Therefore, based on the conversion matrix between the inertial coordinate system and the ground coordinate system, a kinematic model and a dynamic model with input lag of the control signal can be established. For example, the kinematic model is as follows:

[0097]

[0098] The kinetic model is as follows:

[0099]

[0100]

[0101] Among them, H u , H υ , H r is the nonlinear fluid damping term. X1=X u , X2=X |u|u , Y1=Y υ , Y2=y |υ|υ , Y3=Y |r|υ , N1=N r , N2=N |υ|r , N3=N |r|r are the linear and second-order hydrodynamic coefficients of the ship in surge, sway and pitch motion. Where η = [x, y, ψ] T is the position and direction vector of the underactuated surface ship in the inertial reference frame, v = [u, υ, r] T is the velocity and angular velocity vector, M = diag (m u , m v , m r ) is the mass matrix, τ=[τ u , 0, τ r ] T is the propulsion force and torque vector of the underactuated surface vehicle, ω=[ω u ,ω υ ,ω r ] T is the external disturbance vector, T=[T u ,0,T r ] T is the time constant of the unknown bounded input lag, satisfying T≤T m , T m is the upper bound of the input lag. It should be noted that a dot on a letter represents the first-order derivative.

[0102] In some embodiments, the velocity tracking error includes a swell velocity tracking error and a swell velocity tracking error. The dynamic model is converted to obtain a differential equation of the swell velocity tracking error and a differential equation of the swell velocity tracking error as follows:

[0103]

[0104]

[0105] Then, according to the differential equations corresponding to the above-mentioned surge velocity tracking error and surge velocity tracking error, the first-order integral sliding surface s1 and the second-order integral sliding surface s2 are designed as shown below:

[0106]

[0107] Among them, ξ u , υ , a1 and a2 are constant term coefficients, ξ u >0,a1>0,ξ υ >0,a2>0. σ u and σ υ is the filter error.

[0108] Then, based on the differential equation of the above-mentioned surge velocity tracking error and the first-order integral sliding surface s1, the surge control law can be obtained, and based on the differential equation of the sway velocity tracking error and the second-order integral sliding surface x2, the bow roll control law can be obtained.

[0109] In this embodiment, since the actuator of the target object has an input lag in the control signal, which will cause a lagging change in position and posture, a dynamic model with an input lag in the control signal is established, and a driving control law with an input lag is obtained based on the dynamic model and the integral sliding surface. Therefore, a more accurate control signal can be generated by utilizing the driving control law with an input lag.

[0110] In some embodiments, the travel control law includes a surge control law and a yaw control law; the speed tracking error includes a surge speed tracking error and a yaw speed tracking error; the integral sliding surface includes a first-order integral sliding surface corresponding to the surge speed tracking error and a second-order integral sliding surface corresponding to the yaw speed tracking error; based on the integral sliding surface and the differential equation corresponding to the speed tracking error, a travel control law with input time lag is obtained, including: based on the first-order integral sliding surface and the differential equation corresponding to the surge speed tracking error, a surge control law with input time lag is obtained; based on the second-order integral sliding surface and the differential equation corresponding to the yaw speed tracking error, a yaw control law with input time lag is obtained.

[0111] Specifically, after obtaining the first-order integral sliding surface, the first-order integral sliding surface is differentiated, and then the differential equation corresponding to the sway velocity tracking error is substituted into the first-order integral sliding surface after differentiation, and the relationship between the first-order integral sliding mode and the sway control law is obtained, and then the sway control law is obtained using the sliding mode control theory. Then, the second-order integral sliding surface is differentiated, and the second-order derivative of the differential equation corresponding to the sway velocity tracking error is obtained, and the second-order derivative expression corresponding to the sway velocity tracking error is substituted into the second-order integral sliding surface after differentiation, and the relationship between the second-order integral sliding mode and the heading control law is obtained, and then the heading control law is obtained using the sliding mode control theory.

[0112] For example, taking the derivative of the first-order integral sliding surface and substituting the differential equation corresponding to the sloshing velocity tracking error into it, we get:

[0113]

[0114] Among them, f u is a nonlinear term, is the derivative of the filter output.

[0115] Then, according to the sliding mode control theory, the turbulence control law τ is obtained: uAs shown below:

[0116]

[0117] Among them, λ1>0, γ1 is the gain matrix. Yes u An estimate of ε m is a small positive constant that represents the estimation error of the target neural network. W u is the weight parameter of the target neural network.

[0118] Deriving the second-order integral sliding surface and substituting the second-order derivative expression corresponding to the sway velocity tracking error into it, we get:

[0119]

[0120] Where g1 = -m u u(tT υ ). υ is a nonlinear term, is the derivative of the filter output.

[0121] Then, according to the sliding mode control theory, the bow control law τ is obtained: r As shown below:

[0122]

[0123] Among them, λ2>0, γ1 is the gain matrix. Yes v The estimated value of . W υ is the weight parameter of the target neural network.

[0124] In some embodiments, the target neural network has a weight adaptive law for calculating an estimated value of a weight parameter, and the weight adaptive laws corresponding to the first nonlinear term and the second nonlinear term are respectively as follows:

[0125]

[0126]

[0127] Wherein, s1 and s2 represent the first-order integral sliding surface and the second-order integral sliding surface, respectively. N1 and N2 represent the neuron vector calculated based on the longitudinal velocity tracking error and the neuron vector calculated based on the transverse velocity tracking error, respectively. and is the adaptive gain parameter, ∈1 and ∈2 are positive design parameters.

[0128] In some embodiments, before determining the driving control law, it is necessary to assume that the target object satisfies the following assumptions. Assumption 1: The expected position vector η of the target object d is differentiable; Assumption 2: The control signal and speed of the target object are bounded, that is, in is a known upper bound; Assumption 3: The external disturbance ω of the target object is time-varying and bounded, that is, in is an unknown positive constant; Assumption 4: All parameter perturbations have upper bounds.

[0129] In this embodiment, the surge control law with input time lag is obtained based on the differential equation corresponding to the first-order integral sliding surface and the surge velocity tracking error, and the bow roll control law with input time lag is obtained based on the differential equation corresponding to the second-order integral sliding surface and the sway velocity tracking error. The surge control law of the target object is determined, and the velocity error convergence problem of the target object is solved by utilizing the integral sliding mode, thereby achieving the tracking performance of the entire closed-loop system.

[0130] In some embodiments, the target object is an under-actuated surface vessel; after generating the control signal, the method further includes: sending the control signal to the under-actuated surface vessel; the under-actuated surface vessel receives the control signal and travels to a desired position at a second moment according to the control signal.

[0131] Among them, underactuation means that the number of system actuators is less than the number of degrees of freedom. For example, the degrees of freedom of a surface ship include surge, sway and pitch, and the actuators are composed of actuators corresponding to surge and pitch, respectively. That is, the control signal includes surge control signal and pitch control signal. The surge control signal is used to control the longitudinal propulsion force of the underactuated surface ship, and the pitch control signal is used to control the steering propulsion force of the underactuated surface ship.

[0132] Specifically, when the target object is an underactuated surface ship, the controller generates a control signal for the underactuated surface ship and sends the control signal to the actuator of the underactuated surface ship. The actuator of the underactuated surface ship receives the control signal and generates a longitudinal propulsion force and a steering propulsion force that enable the underactuated surface ship to travel to a desired position at the second moment. Figure 4 As shown in the figure, a schematic diagram of trajectory tracking of an underactuated surface ship is shown. The control objective is to make the underactuated surface ship at the actual position (x, y) travel to the desired position (x) at a specified time. d ,y d ).

[0133] In this embodiment, the trajectory tracking control of the underactuated surface vessel is achieved by sending a control signal to the underactuated surface vessel.

[0134] In some embodiments, Lyapunov stability theory can be used to determine that the use of a virtual speed control law can make the position tracking error of the target object converge within a finite time, that is, to prove that the virtual speed control law can perform stability control on the errors in the kinematic model and the dynamic model. For example, based on the differential equation of the position tracking error and the virtual speed control law, an expression for the speed tracking error can be obtained, as shown in the following formula:

[0135]

[0136] When u e ,υ e →0, the above expression has only zero solution, namely:

[0137]

[0138] Select the following Lyapunov function:

[0139]

[0140] Derivative V1, substitute the above differential equation of position tracking error into the derived formula, and we get:

[0141]

[0142] From the above formula, we can see that when u→α u ,υ→α υ When the speed tracking error u e ,u e →0, Guaranteed position tracking error (x e ,y e ) converges to the origin in a finite time, and the proof is complete.

[0143] In some embodiments, the driving control law τ u , τ r Substituting into the differential equation of the integral sliding surface, we get:

[0144]

[0145] According to Barbalat's lemma, consider the function f(t)∈L ∞ , if p∈[1,∞), the function f(t) satisfies f(t)∈L p ,and is bounded, t∈[0,∞), then lim t→∞ f(t)=0. Therefore, t→∞, s1, s2→0, that is, Therefore, the sliding surface convergence with input time lag integral holds. Therefore, the virtual velocity tracking error e u , eυ On this basis, the position tracking error x e ,y e Converges. Therefore, the closed-loop system is globally uniformly stable and ultimately bounded.

[0146] In some embodiments, in order to verify the robustness of the driving control law for the target object to trajectory tracking, model parameter uncertainty and the system with input delay, a simulation is performed on an underactuated surface ship model. Figure 5 The control block diagram of the underactuated surface vessel is shown. Figure 6 The simulation diagram of the control system is shown. The parameters of the underactuated surface ship considered in the simulation are as follows: m 11 =100kg, m 22 =80kg,m 33 =120kg, I z =50kg·m 2 , X u =70kg / s, X u|u| =100kg / s, Y υ =100kg / m, Y υ|υ| =200kg / m, N r =50kg·m 2 / s,N r|r| = =100kg·m 2 / s. And, assuming the input time lag is 0.5s, 0.52s is the maximum limit of the input time lag. In order to simulate the uncertainty of the hydrodynamic coefficients of underactuated surface ships, a 10% parameter perturbation is added to each nominal hydrodynamic coefficient. ω u =3sin(0.4t)N,ω υ =1.2cos(0.3t)N,ω r =4sin(0.25t)N.

[0147] In some embodiments, Figure 7 and Figure 8 The schematic diagram of the change of position tracking error and virtual velocity tracking error with input lag are shown respectively. Considering the initial state value of the system is: x0 = 3m, y0 = 2.5m, ψ0 = -3rad, the control parameter is: λ1 = 3.94×10 -2 ,λ2=3.1475×10 -2 ,γ1=1.5×10 -4 ,γ2=1.5×10 -4The simulation results using the integral sliding mode control law show that the position tracking error quickly converges to the preset range of the performance function, and the velocity tracking error can react quickly to track the desired position. After 10 seconds, the underactuated surface ship achieves the control goal of tracking the predetermined trajectory.

[0148] In some embodiments, Fig. 9 The control force and torque diagram with input lag is shown. The solid line represents the change curve of the control force representing the longitudinal propulsion force, and the dotted line represents the change curve of the torque representing the steering propulsion force. Fig. 9 It can be seen that the response curves of the control force and torque gradually become smooth after 10s, which meets the actual engineering needs.

[0149] In some embodiments, Fig.10 and Fig.11 The following diagrams show the estimated curves of the first nonlinear term representing the composite disturbance and the second nonlinear term representing the composite disturbance by the target neural network. Fig.10 and Fig.11 It can be seen that the estimated curve almost coincides with the actual interference curve, and the estimation error is small.

[0150] In some embodiments, the number of neurons in the hidden layer of the target neural network is 31, the initial values ​​of the weights are all selected to be zero, the basis function widths b1=5, b2=2, b2=4, and the center point Gaussian functions are uniformly distributed between [-8,8].

[0151] It should be understood that, although the steps in the flowcharts involved in the above embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above embodiments may include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0152] Based on the same inventive concept, the embodiment of the present application also provides a target object control device for implementing the target object control method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more target object control device embodiments provided below can refer to the limitations of the target object control method above, and will not be repeated here.

[0153] In some embodiments, Fig.12 As shown, a target object control device is provided, including: a control law determination module 1202, a speed error determination module 1204, a sliding mode calculation module 1206, an estimation compensation module 1208 and a control signal generation module 1210, wherein:

[0154] The control law determination module 1202 is used to determine the driving control law for the target object; the driving control law includes an integral sliding mode term and a nonlinear term.

[0155] The speed error determination module 1204 is used to determine the speed tracking error based on the first position and the first speed of the target object at the first moment and the expected position of the target object corresponding to the second moment; the second moment is the next moment of the first moment.

[0156] The sliding mode calculation module 1206 is used to determine the value of the integral sliding mode term based on the speed tracking error.

[0157] The estimation compensation module 1208 is used to use the target neural network to iteratively estimate the compensation prediction of the nonlinear term based on the speed tracking error. During the estimation compensation prediction process, the number of neurons in the target neural network is iteratively adjusted based on the activation corresponding to the target neural network until the iteration stops, thereby obtaining the estimated value corresponding to the nonlinear term.

[0158] The control signal generating module 1210 is used to substitute the value and estimated value of the integral sliding mode term into the driving control law to generate a control signal; the control signal is used to control the target object to travel to the desired position at the second moment.

[0159] In some embodiments, the estimation compensation module 1208 is also used to: determine the activation degree corresponding to each neuron based on the activation function of the target neural network and the speed tracking error during each round of iteration of the estimation compensation prediction; adjust the number of neurons in the target neural network based on the activation degree of each neuron to obtain an adjusted target neural network; determine the weight parameters corresponding to the adjusted target neural network based on the neuron weights corresponding to each neuron in the adjusted target neural network; stop iteration when the weight parameter reaches the ideal weight parameter, and use the target neural network corresponding to the ideal weight parameter to determine the estimated value corresponding to the nonlinear term based on the speed tracking error.

[0160] In some embodiments, the estimation compensation module 1208 is also used to: determine the target activation from the activations corresponding to each neuron; when the target activation is less than or equal to a first threshold, add new neurons to the target neural network to obtain an adjusted target neural network; wherein the neuron weights corresponding to the new neurons are determined based on the neuron weights corresponding to the neurons to which the target activation belongs.

[0161] In some embodiments, the estimation compensation module 1208 is also used to: for each neuron, calculate the index value corresponding to the neuron based on the activation degree corresponding to the neuron; when the index value of the neuron is less than or equal to the second threshold, delete the neuron to obtain an adjusted target neural network.

[0162] In some embodiments, the speed error determination module 1206 is also used to: determine a virtual speed control law for the target object; substitute the first position and the expected position into the virtual speed control law to obtain the expected speed corresponding to the target object; and determine the speed tracking error based on the difference between the first speed and the expected speed.

[0163] In some embodiments, the control law determination module 1202 is also used to: establish a dynamic model with input time delay of the control signal for the target object; perform model conversion on the dynamic model to obtain a differential equation corresponding to the speed tracking error; the differential equation includes input time delay; determine the integral sliding surface for the speed tracking error; and obtain a driving control law with input time delay based on the integral sliding surface and the differential equation corresponding to the speed tracking error.

[0164] In some embodiments, the driving control law includes a surge control law and a yaw control law; the speed tracking error includes a surge speed tracking error and a yaw speed tracking error; the integral sliding surface includes a first-order integral sliding surface corresponding to the surge speed tracking error and a second-order integral sliding surface corresponding to the yaw speed tracking error; the control law determination module 1202 is also used to: obtain a surge control law with input time lag based on the first-order integral sliding surface and the differential equation corresponding to the surge speed tracking error; obtain a yaw control law with input time lag based on the second-order integral sliding surface and the differential equation corresponding to the yaw speed tracking error.

[0165] In some embodiments, the target object is an under-actuated surface vessel; after generating the control signal, the device is further used to: send the control signal to the under-actuated surface vessel; the under-actuated surface vessel receives the control signal and travels to the desired position at a second moment according to the control signal.

[0166] Each module in the above-mentioned target object control device can be implemented in whole or in part by software, hardware or a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each of the above modules.

[0167] In some embodiments, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Fig.13As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store relevant data involved in the target object control method. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a target object control method is implemented.

[0168] In some embodiments, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Fig.14 As shown. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and the external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a target object control method is implemented. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.

[0169] Those skilled in the art will understand that Fig.13 or Fig.14The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0170] In some embodiments, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned target object control method when executing the computer program.

[0171] In some embodiments, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned target object control method are implemented.

[0172] In some embodiments, a computer program product is provided, including a computer program, which implements the steps in the above-mentioned target object control method when executed by a processor.

[0173] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.

[0174] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0175] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0176] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A target object control method, characterized in that: The method comprises: Determining a driving control law for a target object; the driving control law includes an integral sliding mode term and a nonlinear term; Determining a velocity tracking error based on a first position and a first velocity of the target object at a first moment and an expected position of the target object corresponding to a second moment; the second moment is a moment next to the first moment; Determining a value of the integral sliding mode term based on the speed tracking error; Using a target neural network to iteratively estimate, compensate and predict the nonlinear term based on the speed tracking error, during the estimation, compensation and prediction process, iteratively adjust the number of neurons in the target neural network based on the activation corresponding to the target neural network until the iteration stops, and obtain an estimated value corresponding to the nonlinear term; wherein, during each round of iterative process of estimation, compensation and prediction, the number of neurons in the target neural network is adjusted based on the activation corresponding to each neuron in the target neural network; Substituting the value of the integral sliding mode term and the estimated value into the driving control law to generate a control signal; the control signal is used to control the target object to travel to the desired position at the second moment; The step of adjusting the number of neurons in the target neural network based on the activation degree corresponding to each neuron in the target neural network comprises: For each neuron, using a preset index value algorithm, an index value corresponding to the neuron is calculated based on the activation degree corresponding to the neuron, and when the index value of the neuron is less than or equal to a second threshold, the neuron is deleted to obtain an adjusted target neural network; The preset index value algorithm is: Among them, I r is the index value of the rth neuron, is the index value of the current rth neuron, N r is the activation of the rth neuron, M th and l are positive constants.

2. The method according to claim 1, characterized in that In the estimation compensation prediction process, the number of neurons in the target neural network is iteratively adjusted based on the activation corresponding to the target neural network until the iteration stops, and an estimated value corresponding to the nonlinear term is obtained, including: In each round of iteration of the estimation compensation prediction, the activation degree corresponding to each neuron is determined based on the activation function of the target neural network and the speed tracking error; Adjusting the number of neurons in the target neural network based on the activation degree of each of the neurons to obtain an adjusted target neural network; Determining a weight parameter corresponding to the adjusted target neural network according to the neuron weights corresponding to the neurons in the adjusted target neural network; When the weight parameter reaches the ideal weight parameter, the iteration is stopped, and the estimated value corresponding to the nonlinear term is determined based on the speed tracking error using the target neural network corresponding to the ideal weight parameter.

3. The method according to claim 2, characterized in that The step of adjusting the number of neurons in the target neural network based on the activation degree of each of the neurons to obtain an adjusted target neural network includes: Determining a target activation degree from the activation degrees corresponding to each of the neurons; When the target activation is less than or equal to a first threshold, adding a new neuron to the target neural network to obtain an adjusted target neural network; The neuron weight corresponding to the new neuron is determined based on the neuron weight corresponding to the neuron to which the target activation degree belongs.

4. The method according to claim 1, characterized in that: The method further comprises: If the index value of the neuron is greater than the second threshold, the neuron is retained.

5. The method according to claim 1, characterized in that The determining of the velocity tracking error based on the first position and the first velocity of the target object at the first moment and the expected position corresponding to the target object at the second moment includes: determining a virtual speed control law for the target object; Substituting the first position and the expected position into the virtual speed control law to obtain the expected speed corresponding to the target object; The speed tracking error is determined based on a difference between the first speed and the desired speed.

6. The method according to claim 1, characterized in that The determining of the driving control law for the target object includes: For the target object, a dynamic model with input time lag of a control signal is established; Performing model conversion on the dynamic model to obtain a differential equation corresponding to the speed tracking error; the differential equation includes the input time lag; Determining an integral sliding surface according to the speed tracking error; Based on the differential equation corresponding to the integral sliding surface and the speed tracking error, a driving control law with the input time lag is obtained.

7. The method according to claim 6, characterized in that The driving control law includes a surge control law and a yaw control law; the speed tracking error includes a surge speed tracking error and a sway speed tracking error; the integral sliding surface includes a first-order integral sliding surface corresponding to the surge speed tracking error and a second-order integral sliding surface corresponding to the sway speed tracking error; The driving control law with the input time lag is obtained based on the differential equation corresponding to the integral sliding surface and the speed tracking error, including: Based on the first-order integral sliding surface and the differential equation corresponding to the surge velocity tracking error, a surge control law with the input time lag is obtained; Based on the second-order integral sliding mode surface and the differential equation corresponding to the sway velocity tracking error, a yaw control law with the input time lag is obtained.

8. The method according to claim 1, characterized in that The target object is an underactuated surface ship; After generating the control signal, the method further comprises: sending the control signal to the underactuated surface vessel; The underdriven surface vessel receives the control signal and travels to the desired position at the second time according to the control signal.

9. A target object control device, characterized in that: The device comprises: A control law determination module, used to determine a driving control law for a target object; the driving control law includes an integral sliding mode term and a nonlinear term; A speed error determination module, configured to determine a speed tracking error based on a first position and a first speed of the target object at a first moment, and an expected position of the target object corresponding to a second moment; the second moment is a moment next to the first moment; A sliding mode calculation module, used for determining the value of the integral sliding mode term based on the speed tracking error; An estimation compensation module is used to use a target neural network to iteratively estimate and compensate the nonlinear term based on the speed tracking error. During the estimation and compensation prediction process, the number of neurons in the target neural network is iteratively adjusted based on the activation corresponding to the target neural network until the iteration stops, and an estimated value corresponding to the nonlinear term is obtained; wherein, during each round of iteration of the estimation and compensation prediction, the number of neurons in the target neural network is adjusted based on the activation corresponding to each neuron in the target neural network, and for each neuron, a preset index value algorithm is used to calculate the index value corresponding to the neuron based on the activation corresponding to the neuron, and when the index value of the neuron is less than or equal to a second threshold, the neuron is deleted to obtain an adjusted target neural network; the preset index value algorithm is, Among them, I r is the index value of the rth neuron, is the index value of the current rth neuron, N r is the activation of the rth neuron, M th and l are positive constants; A control signal generating module is used to substitute the value of the integral sliding mode term and the estimated value into the driving control law to generate a control signal; the control signal is used to control the target object to travel to the desired position at the second moment.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

12. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.