A model-free trajectory tracking method and system for an unmanned underwater vehicle

Abstract

The application discloses a model-free unmanned underwater vehicle trajectory tracking method and system, which comprises the following steps: step 1, according to a given expected trajectory and a measured unmanned underwater vehicle position, a tracking error and a tracking error change rate are calculated, and a nonlinear non-singular terminal sliding mode surface is constructed; step 2, according to the unmanned underwater vehicle position, speed and expected trajectory, a dynamic regression matrix is calculated; step 3, according to the sliding mode surface and the dynamic regression matrix, a nonlinear adaptive law is designed, and an upper bound of model uncertainty, a lower bound reciprocal of input saturation inverse scale factor and an estimated value of unknown physical parameters of the unmanned underwater vehicle are obtained; and step 4, according to the estimated value obtained in step 3, a model-free robust adaptive finite time control law is designed in combination with a control law gain coefficient and the dynamic regression matrix. The application can solve the problem of trajectory tracking control of the unmanned underwater vehicle under the conditions of input saturation and model uncertainty and unknown physical parameters.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of unmanned underwater vehicles, in particular to a model-free trajectory tracking method and system for unmanned underwater vehicles. BACKGROUND

[0002] As an intelligent equipment integrating underwater detection, environmental perception and autonomous decision-making functions, unmanned underwater vehicles have been widely used in the fields of marine resource exploration, submarine pipeline maintenance, underwater target reconnaissance and military operations. The trajectory tracking control performance, one of the core technologies, directly affects the task execution accuracy and reliability of the vehicle in complex marine environments.

[0003] Current mainstream control methods usually construct control algorithms based on accurate dynamic models. However, in practical applications, the hydrodynamic parameters of the vehicle are easily affected by factors such as carrier deformation, biofilm accumulation and fluid environment changes, resulting in significant time-varying characteristics of model parameters. In addition, factors such as actuator thrust saturation constraints and external disturbances further increase the design complexity of the control system.

[0004] In the prior art, patent document CN118625841A proposes a trajectory tracking method based on online modeling and model predictive control. This technology extracts features from offline data using fuzzy C-means clustering and establishes a non-parametric model using least squares support vector machines. Then, a model predictive controller is used to achieve trajectory tracking. However, this scheme has the following technical defects: (1) The control performance is highly dependent on the accuracy of online modeling, which can easily lead to tracking instability when the vehicle encounters unmodeled dynamics or sudden disturbances; (2) The input saturation nonlinearity caused by limited propeller output torque is not considered, which may cause integral saturation phenomenon; (3) The asymptotic convergence control strategy cannot meet the stringent requirements of finite-time convergence of tracking error in high dynamic scenarios.

[0005] Another related patent document CN118915475A discloses a trajectory tracking method based on super-spiral sliding mode control, which designs an anti-disturbance control law by establishing an accurate mathematical model. Although this technology can suppress the influence of external disturbances, it has obvious limitations: (1) The control law design requires an accurate mathematical model as a prerequisite, which cannot effectively handle model uncertainties caused by parameter perturbations; (2) No input saturation compensation mechanism is constructed, and the physical constraints of the actuator may violate the reaching condition of the sliding surface; (3) The tracking error can only achieve asymptotic convergence, which cannot meet the needs of time-sensitive tasks such as emergency obstacle avoidance and rapid approach.

[0006] In summary, the prior art generally has three common defects: first, over-reliance on accurate mathematical models leads to insufficient robustness; second, there is a lack of stability guarantee mechanism under input saturation constraints; third, the control convergence speed cannot meet the requirements of finite time performance indicators. These defects seriously restrict the reliable operation capability of unmanned underwater vehicles in dynamic uncertain environments. SUMMARY

[0007] The object of the present application is to provide a model-free trajectory tracking method and system for an unmanned underwater vehicle, to overcome or at least mitigate at least one of the above-mentioned defects of the prior art.

[0008] To achieve the above object, the present application provides a model-free trajectory tracking method for an unmanned underwater vehicle, comprising:

[0009] Step 1: Calculate the tracking error and the rate of change of the tracking error according to the given desired trajectory and the measured pose of the unmanned underwater vehicle, and construct a nonlinear nonsingular terminal sliding mode surface;

[0010] Step 2: Calculate the dynamic regression matrix according to the pose, velocity and desired trajectory of the unmanned underwater vehicle;

[0011] Step 3: Design a nonlinear adaptive law according to the sliding mode surface and the dynamic regression matrix, to obtain the upper bound of the model uncertainty, the lower bound of the inverse of the input saturation reverse scaling factor, and the estimated value of the unknown physical parameters of the unmanned underwater vehicle;

[0012] Step 4: Design a model-free robust adaptive finite-time control law based on the estimated values obtained in Step 3, combined with the control law gain coefficient and the dynamic regression matrix.

[0013] Further, the nonlinear nonsingular terminal sliding mode surface of Step 1 is expressed as S n :

[0014] S n =k1f1(e t )+k2f2(e t )+e tv

[0015] where f1(e t ) and f2(e t ) are intermediate parameters, and are expressed as follows, respectively:

[0016] f1(e)=|e tx |sign(e tx ),|e ty |sign(e ty ),|e tz |sign(e tz ),|e tφ|sign(e tφ ),|e tθ |sign(e tθ ),|e tψ |sign(e tψ )] T

[0017]

[0018] where k1≥0.5, k2≥0, 0<p i <1, | | is absolute value, sign() is sign function.

[0019] Further, the dynamic regression matrix of step 2 is expressed as

[0020] where η(t), ξ(t) are measured pose, velocity respectively, η d (t) is given desired trajectory, is desired pose rate, is composed of physical parameters of the unmanned underwater vehicle, superscript T represents transpose, superscript -1 represents reciprocal, the dot above the parameter represents the derivative of the parameter, χ, M * (η(t)), C * (η(t), ξ(t)), D * (η(t), ξ(t)) are intermediate / process parameters used for simplifying the formula, which are specifically described as follows:

[0021]

[0022] M * (η(t))=J -T (η(t))MJ -1 (η(t))

[0023]

[0024] D * (η(t), ξ(t))=J -T (η(t))D(ξ(t))J -1 (η(t))。

[0025] where J(η(t)), M, C(ξ(t)), D(ξ(t)) are coordinate conversion matrix, inertia matrix, rigid body centripetal force and Coriolis force matrix, and hydrodynamic damping matrix respectively.

[0026] Further, the nonlinear adaptive law of step 3 is expressed as follows:

[0027]

[0028] wherein τ dM is an unknown upper bound of model uncertainty, γ is a lower bound reciprocal of input saturation inverse scale factor, Γ is a set coefficient greater than 0, is a preset gain coefficient greater than 0, the symbol ∧ above a parameter represents an estimated value of the parameter, and the symbol · above a parameter represents a derivative of the parameter, is a gain coefficient, and is represented as follows:

[0029]

[0030] Further, the model-free robust adaptive finite time control law τ (t) of step 4 is represented as follows:

[0031]

[0032] The application also provides a model-free unmanned underwater vehicle trajectory tracking system, which comprises a desired trajectory giving module, a tracking control module and an unmanned underwater vehicle, wherein the unmanned underwater vehicle specifically comprises:

[0033] a nonlinear nonsingular terminal sliding mode surface construction unit, which is configured to calculate a tracking error and a tracking error change rate, and construct a nonlinear nonsingular terminal sliding mode surface according to a given desired trajectory and a measured unmanned underwater vehicle pose;

[0034] a dynamic regression matrix calculation unit, which is configured to calculate a dynamic regression matrix according to the unmanned underwater vehicle pose, velocity and desired trajectory;

[0035] a nonlinear adaptive law design unit, which is configured to design a nonlinear adaptive law according to the sliding mode surface and the dynamic regression matrix, and obtain an unknown upper bound of model uncertainty, a lower bound reciprocal of input saturation inverse scale factor, and an estimated value of an unknown physical parameter of the unmanned underwater vehicle;

[0036] a model-free robust adaptive finite time control law design unit, which is configured to design a model-free robust adaptive finite time control law according to the estimated value of the nonlinear adaptive law design unit, in combination with a control law gain coefficient and the dynamic regression matrix.

[0037] Further, the nonlinear nonsingular terminal sliding mode surface of the nonlinear nonsingular terminal sliding mode surface construction unit is represented as S n :

[0038] S n = k1f1(e t ) + k2f2(e t ) + e tv

[0039] wherein f1(e t ) and f2(et ) are intermediate parameters, which are defined as follows:

[0040] f1(e t ) = [ |e tx | sign(e tx ), |e ty | sign(e ty ), |e tz | sign(e tz ), |e tφ | sign(e tφ ), |e tθ | sign(e tθ ), |e tψ | sign(e tψ )] T

[0041]

[0042] where k1≥0.5, k2≥0, 0 i <1, || is absolute value, and sign() is sign function.

[0043] Further, the dynamic regression matrix of the dynamic regression matrix calculation unit is represented as

[0044]

[0045] where η(t), ξ(t) are measured pose, velocity, respectively, η d (t) is given desired trajectory, is desired pose rate, is composed of physical parameters of the unmanned underwater vehicle, superscript T represents transpose, superscript -1 represents reciprocal, · above the parameter head represents derivative of the parameter, X, M * (η(t)), C * (η(t), ξ(t)), D * (η(t), ξ(t)) are intermediate / process parameters for simplifying formula, which are described as follows:

[0046]

[0047] M * (η(t)) = J -T (η(t)) MJ -1 (η(t))

[0048]

[0049] D *(eta(t),xi(t))=J -T (eta(t))D(xi(t))J -1 (eta(t))。

[0050] Wherein, J(eta(t)), M, C(xi(t)), D(xi(t)) are coordinate conversion matrix, inertia matrix, rigid body centripetal force and Coriolis force matrix, hydrodynamic damping matrix respectively.

[0051] Further, the nonlinear adaptive law of the nonlinear adaptive law design unit is expressed as the following formula:

[0052]

[0053] Wherein, tau dM is the upper bound of model uncertainty, gamma is the lower reciprocal of input saturation inverse scale factor, Gamma is a set coefficient greater than 0, is a preset gain coefficient greater than 0, the parameter above the head represents the estimated value of the parameter, and the parameter above the head represents the derivative of the parameter, is a gain coefficient, and is expressed as follows:

[0054]

[0055] Further, the model-free robust adaptive finite-time control law tau(t) is expressed as the following formula:

[0056]

[0057] The present application has the following advantages due to the above technical scheme:

[0058] The present application can solve the problem of trajectory tracking control of unmanned underwater vehicle physical parameters under the condition of input saturation and model uncertainty, so as to eliminate the dependence on the physical parameters of unmanned underwater vehicle, improve the robustness of input saturation and model uncertainty, and realize the finite-time convergence of tracking error. BRIEF DESCRIPTION OF DRAWINGS

[0059] Figure 1 It is a flow chart of the model-free unmanned underwater vehicle trajectory tracking control method of the embodiment of the present application.

[0060] Figure 2 It is a schematic diagram of the model-free unmanned underwater vehicle trajectory tracking control system according to the embodiment of the present application. DETAILED DESCRIPTION

[0061] In the drawings, the same or similar signs are used to represent the same or similar elements or elements with the same or similar functions. The embodiments of the present application will be described in detail below with reference to the drawings.

[0062] In the description of the present application, the terms "center", "longitudinal", "transverse", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the scope of protection of the present application.

[0063] As shown in Figure 1 The model-free unmanned underwater vehicle trajectory tracking method provided by the embodiments of the present application comprises:

[0064] Step 1, according to the given desired trajectory and the measured unmanned underwater vehicle pose, the tracking error and the tracking error change rate are calculated, and a nonlinear nonsingular terminal sliding mode surface is constructed.

[0065] For example, in one embodiment, the nonlinear nonsingular terminal sliding mode surface of step 1 is represented as S n :

[0066] S n =k1f1(e t )+k2f2(e t )+e tv (1)

[0067] Wherein:

[0068] e t represents the tracking error, e t =[e tx ,e ty ,e tz ,e tφ ,e tθ ,e tψ ] T , (e tx ,e ty ,e tz ), (e tφ ,e tθ ,e tψ ) are the position tracking error, attitude tracking error in three directions of the inertial coordinate system xyz, e t is obtained by formula (2):

[0069] e t =η(t)-η d (t) (2)

[0070] In formula (2), η(t) is the measured pose, η d (t) is the given desired trajectory.

[0071] e tv tracking error rate, e tv = [e tvx , e tvy , e tvz , e tvφ , e tvθ , e tvψ ] T , (e tvx , e tvy , e tz ), (e tvφ , e tvθ , e tvψ ] are the position tracking error rate, the attitude tracking error rate, e tv are obtained by formula (3):

[0072]

[0073] In formula (3), is the pose rate, is the expected pose rate.

[0074] k1 and k2 are preset values, which satisfy k1≥0.5 and k2≥0, and have no substantial physical meaning.

[0075] f1(e t ) and f2(e t ) are intermediate / process parameters with no substantial physical meaning, and are represented as follows:

[0076] f1(e t ) = [|e tx |sign(e tx ), |e ty |sign(e ty ), |e tz |sign(e tz ), |e tφ |sign(e tφ ), |e tθ |sign(e tθ ), |e tψ |sign(e tψ

[0077]

[0078] wherein || is the absolute value, 0 i <1, p i denotes the exponential, which has no substantial physical meaning, and k1, k1 and p iThe value range of these parameter settings is conducive to ensuring the convergence of the subsequent designed nonlinear adaptive law and the finite time stability of the model-free robust adaptive control law. sign(x) is a sign function, which can be implemented by equation (4):

[0079]

[0080] sign(x) can also be represented by a continuous saturation function f s (x) of equation (5) to eliminate the chattering caused by the discontinuous function sign(x):

[0081]

[0082] In equation (5), ρ1 and ρ2 are pre-set values, which satisfy 0 < ρ1 < 1 and ρ2 > 0, and have no substantial physical meaning.

[0083] The nonlinear non-singular terminal sliding mode surface constructed according to the tracking error and the tracking error rate is conducive to eliminating the singularity phenomenon caused by the traditional terminal sliding mode surface, can ensure the finite time stability and robustness of the controller, and can reduce the chattering caused by the sliding mode control.

[0084] Step 2, according to the pose, velocity and desired trajectory of the unmanned underwater vehicle, the dynamic regression matrix is calculated.

[0085] According to equation (1) and equation (3), the nonlinear non-singular terminal sliding mode surface S n can be further represented as Then wherein, χ represents an intermediate / process parameter for simplifying the formula, which has no specific physical meaning.

[0086] According to The Euler-Lagrange motion model of the unmanned underwater vehicle in the inertial coordinate system is represented by equation (6):

[0087]

[0088] wherein, τ d (t) represents the uncertainty parameter of the Euler-Lagrange motion model, ISN(τ(t)) is a control input saturation function and is described by equation (7):

[0089]

[0090] wherein:

[0091] τ M , τ m represent the upper and lower bounds of the uncertainty τ(t) respectively, which are unknown but constant, and satisfy τM >0, and τ m <0.

[0092] M * (η(t)), C * (η(t), ξ(t)), D * (η(t), ξ(t)) represent intermediate / process parameters for simplifying the formula, without specific physical meaning, which are respectively represented as formula (8), (9), (10):

[0093] M * (η(t)) = J -T (η(t)) MJ -1 (η(t)) (8)

[0094]

[0095] D * (η(t), ξ(t)) = J -T (η(t)) D(ξ(t)) J -1 (η(t)) (10)

[0096] Wherein, J(η(t)), M, C(ξ(t)), D(ξ(t)) are respectively coordinate conversion matrix, inertia matrix, rigid body centripetal force and Coriolis force matrix, and hydrodynamic damping matrix.

[0097] The embodiment introduces a reverse scale factor to construct an input saturation function for processing input saturation caused by physical constraints of the thruster and improving the smoothness of the control input.

[0098] According to the Euler-Lagrange motion model of the unmanned underwater vehicle, a dynamic regression matrix is obtained through parameter linearization, as shown in formula (11)

[0099]

[0100] In formula (11), η(t) and ξ(t) are respectively measured pose and velocity, is the expected pose rate of change, η d (t) is a given expected trajectory, the superscript -1 represents the inverse, and the · above the parameter head represents the derivative of the parameter, which is composed of physical parameters of the unmanned underwater vehicle, such as mass, moment of inertia, hydrodynamic coefficients, etc.

[0101] When the physical parameters of the unmanned underwater vehicle are unknown, considering the adverse effects of input saturation and model uncertainty, we first construct the motion model of the unmanned underwater vehicle in the body coordinate system, and then transform it into the Euler-Lagrange motion model in the inertial coordinate system. Then, we combine the nonlinear nonsingular terminal sliding surface with the Euler-Lagrange motion model and obtain the dynamic regression matrix of the linear regression model through parameter linearization.

[0102] Step 3: Based on the sliding surface and dynamic regression matrix, design a nonlinear adaptive law to obtain the upper bound of the unknown model uncertainty, the reciprocal of the lower bound of the input saturation inverse scaling factor, and the estimated values ​​of the unknown physical parameters of the unmanned underwater vehicle.

[0103] In one embodiment, the nonlinear adaptive law is expressed as Equation (12), which is used to estimate the unknown upper bound τ of the model uncertainty. dM Input saturation inverse scaling factor lower bound γ and unknown physical parameter vector of the unmanned underwater vehicle

[0104]

[0105] In equation (12), τ dM Let Γ be the upper bound of the unknown model uncertainty, γ be the reciprocal of the lower bound of the input saturation inverse scaling factor, and Γ be a set coefficient greater than 0. The parameter is a preset gain coefficient greater than 0. The ∧ symbol above the parameter indicates the estimated value of the parameter, and the · symbol above the parameter indicates the derivative of the parameter. The gain coefficient is expressed as equation (13):

[0106]

[0107] Therefore, although the physical parameters of an unmanned underwater vehicle are affected by input saturation and model uncertainty, they can still be estimated using the aforementioned nonlinear adaptive law.

[0108] By designing a nonlinear adaptive law based on the sliding mode surface and dynamic regression matrix to estimate the upper bound of the unknown uncertainty of the model, the lower bound of the input saturation inverse scaling factor, and the unknown physical parameters of the unmanned underwater vehicle, the design of the control law does not require prior knowledge of the upper bound of the model uncertainty, the lower bound of the input saturation, and the physical parameters of the unmanned underwater vehicle, thus avoiding dependence on these key parameters of the unmanned underwater vehicle.

[0109] Step 4: Based on the estimated values ​​obtained in Step 3, and combined with the control law gain coefficient and dynamic regression matrix, design a model-free robust adaptive finite-time control law.

[0110] In one embodiment, the model-free robust adaptive finite-time control law τ(t) is expressed as equation (14):

[0111]

[0112] According to the model-free robust adaptive finite time control law t(t) described above, the model-free robust adaptive finite time trajectory tracking control of the unmanned underwater vehicle under the conditions of input saturation and model uncertainty and unknown physical parameters can be realized, without the prior knowledge of the physical parameters of the unmanned underwater vehicle, which is beneficial to guarantee the tracking accuracy, robustness and finite time stability of the system.

[0113] As shown in Figure 2 The embodiment of the present application also provides a model-free unmanned underwater vehicle trajectory tracking system, which comprises an expected trajectory giving module, a tracking control module and an unmanned underwater vehicle, wherein:

[0114] The expected trajectory giving module is used to give the expected trajectory that the unmanned underwater vehicle needs to track, and the unmanned underwater vehicle is a controlled object.

[0115] The tracking control module comprises a nonlinear nonsingular terminal sliding mode surface construction unit, a dynamic regression matrix calculation unit, a nonlinear adaptive law design unit and a model-free robust adaptive finite time control law design unit.

[0116] The nonlinear nonsingular terminal sliding mode surface construction unit is used to calculate the tracking error and the tracking error rate of change according to the given expected trajectory and the measured unmanned underwater vehicle pose, and to construct a nonlinear nonsingular terminal sliding mode surface.

[0117] The dynamic regression matrix calculation unit is used to calculate a dynamic regression matrix according to the unmanned underwater vehicle pose, speed and expected trajectory.

[0118] The nonlinear adaptive law design unit is used to design a nonlinear adaptive law according to the sliding mode surface and the dynamic regression matrix, to obtain the upper bound of the model uncertainty, the lower bound inverse of the input saturation reverse scaling factor, and the estimated value of the unknown physical parameters of the unmanned underwater vehicle.

[0119] The model-free robust adaptive finite time control law design unit is used to design a model-free robust adaptive finite time control law according to the estimated value of the nonlinear adaptive law design unit, in combination with the control law gain coefficient and the dynamic regression matrix.

[0120] In one embodiment, the nonlinear nonsingular terminal sliding mode surface of the nonlinear nonsingular terminal sliding mode surface construction unit is represented as S n .

[0121] In one embodiment, the dynamic regression matrix of the dynamic regression matrix calculation unit is represented as

[0122] In one embodiment, the nonlinear adaptive law of the nonlinear adaptive law design unit is represented as equation (12).

[0123] In one embodiment, the model-free robust adaptive finite-time control law τ(t) is represented as equation (14).

[0124] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them. Those skilled in the art should understand that the technical solutions described in the foregoing embodiments can be modified, or some technical features can be replaced by equivalents; these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A model-free unmanned underwater vehicle trajectory tracking method, characterized in that, Comprise: Step 1, according to the given desired trajectory and measured unmanned underwater vehicle pose, the tracking error and tracking error rate of change, construct nonlinear nonsingular terminal sliding mode surface; Step 2, according to the unmanned underwater vehicle pose, speed and desired trajectory, calculate the dynamic regression matrix; Step 3, according to the sliding mode surface and dynamic regression matrix, design nonlinear adaptive law, obtain the upper bound of model uncertainty, the lower bound of input saturation inverse scale factor reciprocal and the estimate of unknown physical parameters of unmanned underwater vehicle; Step 4, according to the estimate obtained in step 3, combined with the control law gain coefficient and dynamic regression matrix, design model-free robust adaptive finite time control law; The dynamic regression matrix of Step 2 is expressed as where η(t), ξ(t) are measured pose, velocity, respectively, η d (t) is the given desired trajectory, is the desired pose rate of change, is composed of physical parameters of the unmanned underwater vehicle, superscript T denotes transpose, superscript -1 denotes inverse, the dot above the parameter denotes derivative of the parameter, χ, M * (η(t)), C * (η(t)), ξ(t)), D * (η(t)), ξ(t)) are all intermediate / process parameters used for simplifying the formula, which are described as follows: M * (η(t)) = J -T (η(t)) MJ -1 (η(t)) Wherein, J(η(t)), M, C(ξ(t)), D(ξ(t)) are coordinate transformation matrix, inertia matrix, rigid body centripetal force and Coriolis force matrix, hydrodynamic damping matrix respectively; The nonlinear adaptive law of step 3 is expressed as follows: where S n is the nonlinear nonsingular terminal sliding mode surface of Step 1, τ dM is the unknown upper bound of model uncertainty, γ is the lower reciprocal of input saturation inverse scaling factor, Γ is a positive set coefficient, is a positive preset gain coefficient, ∧ above a parameter indicates an estimated value of the parameter, and · above a parameter indicates a derivative of the parameter, is a gain coefficient, and is expressed as follows:

2. The model-free unmanned underwater vehicle trajectory tracking method of claim 1, wherein, S n = klfle t ) + k2f2(e t ) + e tv wherein f1(e t ) and f2(e t ) are intermediate parameters, respectively represented as follows: f1(e t ) = [ |e tx | sign(e tx ), |e ty | sign(e ty ), |e tz | sign(e tz ), |e tφ | sign(e tφ ), |e tθ | sign(e tθ ), |e tψ | sign(e tψ )] T wherein e t represents a tracking error, e t = [e tx , e ty , e tz , e tφ , e tθ , e tψ ] T , (e tx , e ty , e tz ), (e tφ , e tθ , e tψ ) are position tracking errors, attitude tracking errors in three directions of the inertial coordinate system xyz, respectively, e tv represents a tracking error rate of the tracking error, e tv = [e tvx , e tvy , e tvz , e tvφ , e tvθ , e tvψ ] T , (e tvx , e tvy , e tz ), (e tvφ , e tvθ , e tvψ ) are position tracking error rates, attitude tracking error rates in three directions of the inertial coordinate system xyz, respectively, k1≥0.5, k2≥0, 0 i <1, || is an absolute value, and sign() is a sign function.

3. The model-free UUV trajectory tracking method of claim 1 or 2, wherein, The model-free robust adaptive finite time control law τ(t) of step 4 is expressed as follows:

4. A model-free unmanned underwater vehicle trajectory tracking system, characterized by, Comprise desired trajectory given module, tracking control module and unmanned underwater vehicle, wherein the unmanned underwater vehicle specifically comprises: Nonlinear nonsingular terminal sliding mode surface construction unit, which is used to calculate the tracking error and tracking error rate of change according to the given desired trajectory and measured unmanned underwater vehicle pose, and construct nonlinear nonsingular terminal sliding mode surface; Dynamic regression matrix calculation unit, which is used to calculate the dynamic regression matrix according to the unmanned underwater vehicle pose, speed and desired trajectory; Nonlinear adaptive law design unit, which is used to design nonlinear adaptive law according to the sliding mode surface and dynamic regression matrix, and obtain the upper bound of model uncertainty, the lower bound of input saturation inverse scale factor reciprocal and the estimate of unknown physical parameters of unmanned underwater vehicle; Model-free robust adaptive finite time control law design unit, which is used to design model-free robust adaptive finite time control law according to the estimate of nonlinear adaptive law design unit, combined with the control law gain coefficient and dynamic regression matrix; The dynamic regression matrix of the dynamic regression matrix calculation unit is represented as where η(t), ξ(t) are measured pose, velocity, respectively, η d (t) is a given desired trajectory, is the desired pose rate of change, is composed of physical parameters of the unmanned underwater vehicle, superscript T denotes transpose, superscript -1 denotes inverse, the • on top of a parameter denotes derivative of the parameter, χ, M * (η(t)), C * (η(t), ξ(t)), D * (η(t), ξ(t)) are all intermediate / process parameters used for simplifying the formula, which are described as follows: M * (η(t)) = J -T (η(t)) MJ -1 (η(t)) where S n is a nonlinear nonsingular terminal sliding mode surface, J(η(t)), M, C(ξ(t)), D(ξ(t)) are coordinate transformation matrix, inertia matrix, matrix of rigid body centripetal force and Coriolis force, hydrodynamic damping matrix, respectively. The nonlinear adaptive law of nonlinear adaptive law design unit is expressed as follows: where τ dM is an unknown upper bound of the model uncertainty, γ is the inverse of a lower bound of the input saturation reverse scaling factor, Γ is a positive set coefficient, is a positive preset gain coefficient, the ∧ above a parameter represents an estimated value of the parameter, and the • above a parameter represents a derivative of the parameter, is a gain coefficient, and is expressed as follows:

5. The model-free unmanned underwater vehicle trajectory tracking system of claim 4, wherein, S n = klfle t ) + k2f2(e t ) + e tv wherein f1(e t ) and f2(e t ) are intermediate parameters, respectively represented as follows: f1(e t ) = [ |e tx | sign(e tx ), |e ty | sign(e ty ), |e tz | sign(e tz ), |e tφ | sign(e tφ ), |e tθ | sign(e tθ ), |e tψ | sign(e tψ )] T wherein e t represents a tracking error, e t = [e tx , e ty , e tz , e tφ , e tθ , e tψ ] T , (e tx , e ty , e tz ), (e tφ , e tθ , e tψ ) are position tracking errors, attitude tracking errors in three directions of the inertial coordinate system xyz, respectively, e tv represents a tracking error rate of the tracking error, e tv = [e tvx , e tvy , e tvz , e tvψ , e tvθ , e tvψ ] T , (e tvx , e tvy , e tz ), (e tvφ , e tvθ , e tvψ ) are position tracking error rates, attitude tracking error rates in three directions of the inertial coordinate system xyz, respectively, k1≥0.5, k2≥0, 0 i <1, | | is an absolute value, and sign( ) is a sign function.

6. The model-free unmanned underwater vehicle trajectory tracking system of claim 4 or 5, wherein, The model-free robust adaptive finite time control law τ(t) is expressed as follows:

Images