Self-adaptive global fractional order terminal sliding mode snakelike robot control method

Abstract

The invention discloses a self-adaptive global fractional order terminal sliding mode snakelike robot control method. In order to solve the problems of high complexity and high solving difficulty in the modeling process, an error dynamic model is constructed, and the design difficulty of the control system is further simplified. A radial basis function neural network observer is introduced to cope with uncertain items and unknown disturbance possibly existing in the model, and approximate estimation is effectively carried out on the uncertain factors. In order to ensure that the joint snakelike robot system can converge quickly, a self-adaptive constant-speed approach rate control strategy is designed, and a global fractional order terminal sliding mode surface is combined, so that high-precision control of the system is realized, and the robustness of the system is also remarkably enhanced.

Description

Technical Field

[0001] The invention relates to the technical field of electrical automation, and in particular to an adaptive global fractional-order terminal sliding mode snake-like robot control method. Background Art

[0002] Bionic snake-like robots, due to their unique body structure and movement, have shown wide application potential in many fields including land, sea and air. In the field of power grid inspection, when drones cannot perform line inspection tasks in no-fly scenarios, or densely wooded areas are not suitable for manual inspection, snake-like robots have become an ideal alternative with their excellent adaptability to complex terrain and maneuverability in narrow spaces. With improved inspection accuracy, reduced human errors, excellent autonomy and remote control functions, as well as standardized operation and data management capabilities, they provide solid support for the safe and stable operation of the power grid.

[0003] However, when we put this advanced robotics technology into practice, a major challenge we face is that the modeling process of the bionic snake-like robot is extremely complex and difficult to solve, which brings significant difficulties to the design of its control algorithm. Specifically, an overly complex control algorithm may be unable to cope with real-time tracking control, while an overly simple algorithm may not be able to meet the high-precision control requirements. Therefore, it is particularly important to optimize the model of the bionic snake-like robot and develop a more accurate control algorithm to improve its control performance in power grid patrol. Summary of the invention

[0004] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0005] In view of the above existing problems, the present invention is proposed.

[0006] Therefore, the present invention provides an adaptive global fractional-order terminal sliding mode control method for a snake-like robot, and proposes an adaptive global fractional-order terminal sliding mode control method based on a neural network observer while reconstructing the mathematical modeling.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions: including, based on the dynamic model of the N-joint snake-like robot, reconstructing the tracking error dynamic model of the N-joint snake-like robot; constructing a radial basis neural network observer using a cubic B-spline basis function as an activation function to observe uncertainties and external disturbances in the tracking error dynamic model; constructing a global fractional-order terminal sliding surface according to the characteristics of the terminal sliding mode finite time arrival, the global sliding mode fast arrival sliding surface and the dynamic performance of the fractional-order precise control; designing an adaptive constant-speed approach rate based on the global fractional-order terminal sliding surface; combining the constructed radial basis neural network observer, the global fractional-order terminal sliding surface and the adaptive constant-speed approach rate, designing an adaptive global fractional-order terminal sliding mode controller based on a neural network observer for the N-joint snake-like robot system.

[0008] As a preferred solution of the adaptive global fractional-order terminal sliding mode snake robot control method described in the present invention, the tracking error dynamics model of the N-joint snake robot is reconstructed based on the dynamics model of the N-joint snake robot, including:

[0009] The dynamic model of the N-joint snake-like robot is as follows:

[0010]

[0011] Among them, q φ =[φ 1 ,…,φ N-1 ,φ N ,p x ,p y ] T ∈R (N+2)×1 ,φ i is the i-th joint angle of the snake-like robot, (p x ,p y ) is the center of mass coordinate of the snake robot; M(φ)∈R (N+2)×(N+2) is the inertia matrix of the system, is the Coriolis and centripetal force matrix, G(φ)∈R (N+2)×2N and are the gravity and friction matrices, u is the joint input torque, u=[u 1 ,…,u N-1 ,0,0,0] T ∈R (N+2)×1 .

[0012] The tracking error of the N-joint snake robot is defined as:

[0013]

[0014] in, is the target trajectory of the N-joint snake-like robot; q φ (t) is the actual trajectory of the N-joint snake-like robot; e(t)∈R (N +2)×1 is the tracking error of the N-joint snake-like robot.

[0015] The second-order differential of the tracking error of the N-joint snake-like robot is:

[0016]

[0017] in, is the second-order differential of e(t), yes The second-order differential of Yes φ The second-order differential of (t).

[0018] definition Matrix, the dynamic model of the N-joint snake-like robot is changed as follows:

[0019]

[0020] in, is the dimensionless parameter gain matrix,

[0021] Substitute the second-order differential of the tracking error to construct the tracking error dynamics model of the N-joint snake-like robot:

[0022]

[0023] As a preferred solution of the adaptive global fractional-order terminal sliding mode snake robot control method described in the present invention, wherein: a radial basis neural network observer is constructed using a cubic B-spline basis function as an activation function to observe uncertain terms and external disturbances in a tracking error dynamics model, including:

[0024] The radial basis neural network observer is defined as follows:

[0025]

[0026] in, is all the unknown and uncertain terms in the tracking error dynamics model of the N-joint snake-like robot F = W * The approximate value of h(x)+Γ, where h(x) is the hidden layer activation function of the neural network observer. is the optimal weight of the neural network from the hidden layer to the output layer, Γ is between F and The approximate error of It is W * The estimated value of .

[0027] The hidden layer activation function is chosen as the cubic B-spline basis function, which is defined as:

[0028]

[0029] When || xo i ||≤h i When i=1,2…,N+2

[0030]

[0031] When h i <||xo i ||≤2h i When i=1,2…,N+2

[0032]

[0033] When 2h i <||xo i ||, and i=1,2…,N+2

[0034] h i (x) = 0;

[0035] Among them, ||xo i || is the distance between the input vector and the center vector in Euclidean form; x is the input vector; o and o i is the center vector; h i is the width of the cubic B-spline function.

[0036] As a preferred solution of the adaptive global fractional-order terminal sliding mode snake-like robot control method described in the present invention, the global fractional-order terminal sliding mode surface is constructed according to the characteristics of the terminal sliding mode finite time arrival, the global sliding mode fast arrival sliding mode surface and the fractional-order precise control dynamic performance, including:

[0037] The global fractional-order terminal sliding surface is designed as:

[0038]

[0040] Among them, ηe(t) is the tracking error proportional term of the N-joint snake-like robot, and η is the proportional term parameter adjustment gain; is the fractional-order integral term, Υ is the fractional-order term parameter gain, D ε is the fractional integral, and ε is the fractional integral order.

[0041] The first-order differential of the global fractional-order terminal sliding surface is as follows:

[0042]

[0043] in, is the first-order differential of s(t) with respect to time t, They are 1 (t),s 2 (t),…,s N (t),s N+1 (t),s N+2 (t)The first-order differential with respect to time t.

[0044] As a preferred solution of the adaptive global fractional-order terminal sliding mode snake robot control method described in the present invention, wherein: based on the global fractional-order terminal sliding mode surface, an adaptive constant velocity approach rate is designed, including:

[0045] The adaptive constant speed approach rate is designed as follows:

[0046]

[0047] in, is the adaptive term parameter of the constant velocity approach rate,

[0048] The adaptive rate is designed as:

[0049]

[0050] Where ζ is the adaptive rate parameter gain, ζ = diag(ζ 1 ,ζ 2 ,…,ζ N ,ζ N+1 ,ζ N+2 ).

[0051] As a preferred solution of the adaptive global fractional-order terminal sliding mode snake-like robot control method described in the present invention, a controller is designed, including:

[0052] Combining the constructed radial basis neural network observer, the global fractional-order terminal sliding mode surface and the adaptive constant velocity approach rate, the adaptive global fractional-order terminal sliding mode controller of the N-joint snake robot system based on the neural network observer is designed as follows:

[0053]

[0054] The beneficial effects of the present invention are as follows: the present invention optimizes the dynamic model of an N-joint snake-like robot, and solves the problem of extremely complex modeling process and high difficulty in solving by constructing an error dynamics model, thereby facilitating the design of a control algorithm; a radial basis neural network observer is used to perform approximate estimation of uncertainties and unknown disturbances in the error dynamics model; the design of an adaptive constant velocity approach rate and a global fractional-order terminal sliding surface ensures rapid convergence of the jointed snake-like robot system, thereby achieving high-precision control of the system and enhancing robust performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0056] Figure 1 A schematic diagram of the principle control framework of an adaptive global fractional-order terminal sliding mode snake-like robot control method of the present invention;

[0057] Figure 2 This is a comparison chart of the tracking effects of the adaptive global fractional-order terminal sliding mode control method (SMC) based on a neural network observer and the traditional PID control method (PID) for the joint angle 1 of a three-degree-of-freedom snake-like robot at the target trajectory 1 (aim1);

[0058] Figure 3 This is a comparison chart of the tracking effects of the adaptive global fractional-order terminal sliding mode control method (SMC) based on a neural network observer and the traditional PID control method (PID) for the joint angle 2 of the three-degree-of-freedom snake robot under the target trajectory 2 (aim2);

[0059] Figure 4 This is a flowchart of an adaptive global fractional-order terminal sliding mode snake-like robot control method of the present invention. DETAILED DESCRIPTION

[0060] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0061] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0062] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0063] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0064] At the same time, in the description of the present invention, it should be noted that the directions or positional relationships indicated by the terms "upper, lower, inner and outer" are based on the directions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as limiting the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0065] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0066] Example 1

[0067] Reference Figure 1 , Figure 4The first embodiment of the present invention provides an adaptive global fractional-order terminal sliding mode serpentine robot control method. The method of the present invention is based on the dynamic model of the N-joint serpentine robot, reconstructs the tracking error dynamic model of the N-joint serpentine robot; uses the cubic B-spline basis function as the activation function to construct a radial basis neural network observer to observe the uncertainties and external disturbances in the tracking error dynamic model; constructs a global fractional-order terminal sliding surface based on the characteristics of the terminal sliding mode finite time arrival, the global sliding mode fast arrival sliding surface and the dynamic performance of the fractional-order precise control; based on the global fractional-order terminal sliding surface, designs an adaptive constant-speed approach rate; combines the constructed radial basis neural network observer, the global fractional-order terminal sliding surface and the adaptive constant-speed approach rate to design an adaptive global fractional-order terminal sliding mode controller for the N-joint serpentine robot system based on the neural network observer.

[0068] S1. Based on the dynamic model of the N-joint snake-like robot, the tracking error dynamic model of the N-joint snake-like robot is reconstructed. The following points need to be explained in this step:

[0069] The dynamic model of the N-joint snake-like robot is as follows:

[0070]

[0071] Among them, q φ =[φ 1 ,…,φ N-1 ,φ N ,p x ,p y ] T ∈R (N+2)×1 ,φ i is the i-th joint angle of the snake-like robot, (p x ,p y ) is the center of mass coordinate of the snake robot; M(φ)∈ R(N+2)×(N+2) is the inertia matrix of the system, is the Coriolis and centripetal force matrix, G(φ)∈R (N+2)×2N and are the gravity and friction matrices, u is the joint input torque, u=[u 1 ,…,u N-1 ,0,0,0] T ∈R (N+2)×1 .

[0072] The tracking error of the N-joint snake robot is defined as:

[0073]

[0074] in, is the target trajectory of the N-joint snake-like robot; q φ (t) is the actual trajectory of the N-joint snake-like robot; e(t)∈R (N +2)×1 is the tracking error of the N-joint snake-like robot.

[0075] The second-order differential of the tracking error of the N-joint snake-like robot is:

[0076]

[0077] in, is the second-order differential of e(t), yes The second-order differential of Yes φ The second-order differential of (t).

[0078] definition Matrix, the dynamic model of the N-joint snake-like robot is changed as follows:

[0079]

[0080] in, is the dimensionless parameter gain matrix,

[0081] Substitute the second-order differential of the tracking error to construct the tracking error dynamics model of the N-joint snake-like robot:

[0082]

[0083] S2. Use the cubic B-spline basis function as the activation function to construct a radial basis neural network observer to observe the uncertainties and external disturbances in the tracking error dynamics model. What needs to be explained in this step is:

[0084] The radial basis neural network observer is defined as follows:

[0085]

[0086] in, is all the unknown and uncertain terms in the tracking error dynamics model of the N-joint snake-like robot F = W * The approximate value of h(x)+Γ, where h(x) is the hidden layer activation function of the neural network observer. is the optimal weight of the neural network from the hidden layer to the output layer, Γ is between F and The approximate error of It is W* The estimated value of .

[0087] The hidden layer activation function is chosen as the cubic B-spline basis function, which is defined as:

[0088] h(x)=[h 1 (t),h 2 (t),…,h N (t),h N+1 (t),h N+2 (t)]

[0089]

[0090] When || xo i ||≤h i When i=1,2…,N+2

[0091]

[0092] When h i <||xo i ||≤2h i When i=1,2…,N+2

[0093]

[0094] When 2h i <||xo i ||, and i=1,2…,N+2

[0095] h i (x) = 0;

[0096] Among them, ||xo i || is the distance between the input vector and the center vector in Euclidean form; x is the input vector; o and o i is the center vector; h i is the width of the cubic B-spline function.

[0097] S3. According to the characteristics of the terminal sliding mode reaching the sliding mode in finite time, the global sliding mode reaching the sliding mode surface quickly and the dynamic performance of the fractional-order precise control, the global fractional-order terminal sliding mode surface is constructed. The following points need to be explained in this step:

[0098] The global fractional-order terminal sliding surface is designed as:

[0099]

[0100]

[0101] Among them, ηe(t) is the tracking error proportional term of the N-joint snake-like robot, and η is the proportional term parameter adjustment gain; is the fractional-order integral term, Υ is the fractional-order term parameter gain, D ε is the fractional integral, and ε is the fractional integral order.

[0102] The first-order differential of the global fractional-order terminal sliding surface is as follows:

[0103]

[0104] in, is the first-order differential of s(t) with respect to time t, They are 1 (t),s 2 (t),…,s N (t),s N+1 (t),s N+2 (t)The first-order differential with respect to time t.

[0105] S4. Based on the global fractional-order terminal sliding surface, design an adaptive constant velocity approach rate. The following should be noted in this step:

[0106] The adaptive constant speed approach rate is designed as follows:

[0107]

[0108]

[0109] in, is the adaptive term parameter of the constant velocity approach rate,

[0110] The adaptive rate is designed as:

[0111]

[0112] Where ζ is the adaptive rate parameter gain, ζ = diag(ζ 1 ,ζ 2 ,…,ζ N ,ζ N+1 ,ζ N+2 ).

[0113] S5. Combining the constructed radial basis neural network observer, the global fractional-order terminal sliding mode surface and the adaptive constant velocity approach rate, an adaptive global fractional-order terminal sliding mode controller based on a neural network observer for the N-joint snake-like robot system is designed. The following should be noted in this step:

[0114] The controller is designed to:

[0115]

[0116] Preferably, the present embodiment also needs to explain that, compared with the prior art, the present embodiment discloses an adaptive global fractional-order terminal sliding mode snake robot control method, firstly, an error dynamics model is designed to reconstruct the existing dynamics model of the snake robot, and the dynamics model is simplified to solve the problem that the modeling process is extremely complex and difficult to solve; a radial basis neural network observer is used to deal with the uncertainty inside the system and the interference of the external environment, and the uncertainty and disturbance of the system are estimated in real time and accurately by relying on the powerful approximation ability of the radial basis neural network observer, thereby improving the robustness of the system. The design of the global fractional-order terminal sliding mode surface ensures the rapid convergence of the system, and the combination of the adaptive constant velocity approach rate ensures the improvement of the system control accuracy, thereby achieving the improvement of the overall control performance.

[0117] Example 2

[0118] Reference Figure 2-3 , which is another embodiment of the present invention. This embodiment is different from the first embodiment in that it provides a test verification of an adaptive global fractional-order terminal sliding mode snake-like robot control method, including:

[0119] In order to verify and illustrate the technical effect adopted in this method, in this embodiment, a traditional PID control method is selected to be compared with the method of the present invention, and the test results are compared by scientific demonstration to verify the real effect of this method.

[0120] The physical parameters of the three-degree-of-freedom snake-like robot are: the joint links of the snake-like robot N = 4, the mass of a single link of the snake-like robot m = 0.25 kg and the mass of each link is evenly distributed, the joint length l = 0.08 m, and the ground friction c t =1, c n =10, J = 0.0064 kg × m 2 .

[0121] The vector and inertia matrix used by the three-degree-of-freedom snake robot system are: A =

[0122]

[0123] The input and output vectors of the snake robot are The initial conditions of the snake robot are

[0124] The target trajectory of joint angle 1 is: The target trajectory of joint angle 2 is: ref2 =sin(πt).

[0125] The parameters of the adaptive global fractional-order terminal sliding mode controller based on the neural network observer of the present invention are: η=diag(0.85,4.75,1.5,2.7,10.3), γ=diag(7.35,4.2,2.5,0.85,3.52), ε=1.3, ζ=diag(0.75,2.3,1.235,7.5,2.38), h i =2.3,5.8,4.2,1.6,9.4],

[0126] The comparison controller is a traditional PID controller: The comparison controller parameter is k tp =diag(3.5,2.95,0,0,0), k ti =diag(10.25,8.35,0,0,0),k td =diag(0.02.0.015,0,0,0).

[0127] Test environment: Reference Figure 1 In the SIMULINK environment of MATLAB 2019a, a snake robot controlled object model is built according to the existing snake robot physical quantity parameters and mathematical models. The adaptive global fractional-order terminal sliding mode control method based on the neural network observer and the traditional PID control method are used to track and control the drivable joint 1 and drivable joint 2 under different target trajectories and obtain the test result data. Both methods will start the automated test equipment and use MATLB software programming to implement the simulation test of the comparison method. The simulation data is obtained according to the test results. Each method tests 4 groups of data, and each group of data is sampled for 15s. The target trajectory of each group of data input and the tracking trajectory of the output are calculated and compared to verify the feasibility of the proposed algorithm.

[0128] Reference Figure 2 The present invention compares the tracking effect of the adaptive global fractional-order terminal sliding mode control method (SMC) based on the neural network observer and the traditional PID control method (PID) under the target trajectory 1 (aim1). Figure 3The present invention can drive the joint angle 2 under the target trajectory 2 (aim2) to track the adaptive global fractional order terminal sliding mode control method (SMC) based on the neural network observer and the traditional PID control method (PID) under the target trajectory 2. As can be seen from the figure, whether it is for the target trajectory 1 (aim1) or the target trajectory 2 (aim2), the adaptive global fractional order terminal sliding mode control method (SMC) based on the neural network observer and the traditional PID control method (PID) can track the target in a limited time. However, compared from multiple dimensions such as convergence speed, steady-state error and stability, the adaptive global fractional order terminal sliding mode control method (SMC) based on the neural network observer is significantly better than the traditional PID control method (PID). This advantage is mainly due to the real-time compensation ability of the radial basis neural network observer for external disturbances and uncertainties, thereby enhancing the robust performance of the system, and the characteristics of the designed adaptive global fractional order terminal sliding mode controller, which can ensure global convergence and reach the sliding surface in a limited time, achieving rapid convergence while ensuring the control accuracy and stability of the system.

[0129] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An adaptive global fractional-order terminal sliding mode serpentine inspection robot control method, characterized by: include, S1. Based on the dynamic model of the N-joint snake-like robot, the tracking error dynamic model of the N-joint snake-like robot is reconstructed; S2, using cubic B-spline basis function as activation function to construct radial basis neural network observer to observe uncertainties and external disturbances in the tracking error dynamics model; S3, constructing a global fractional-order terminal sliding surface based on the characteristics of the terminal sliding mode finite time arrival, the global sliding mode fast arrival sliding surface and the fractional-order precise control dynamic performance; S4. Design an adaptive constant velocity approach rate based on the global fractional-order terminal sliding surface; S5. Combining the constructed radial basis neural network observer, the global fractional-order terminal sliding mode surface and the adaptive constant velocity approach rate, an adaptive global fractional-order terminal sliding mode controller based on a neural network observer is designed for an N-joint snake-like robot system.

2. The adaptive global fractional-order terminal sliding mode snake robot control method according to claim 1, characterized in that: The dynamic model of the S1 N-joint snake-like robot is as follows, Among them, q φ =[φ1,…,φ N-1 ,φ N ,p x ,p y ] T ∈R (N+2)×1 ,φ i is the i-th joint angle of the snake-like robot, (p x ,p y ) is the center of mass coordinate of the snake robot; M(φ)∈ R(N+2)×(N+2) is the inertia matrix of the system, is the Coriolis and centripetal force matrix, and are the gravity and friction matrices, u is the joint input torque, u=u1,…,u N-1 ,0,0,0] T ∈R (N+2)×1 ; The tracking error of the N-joint snake robot is defined as: in, is the target trajectory of the N-joint snake-like robot; q φ (t) is the actual trajectory of the N-joint snake-like robot; e(t)∈R (N+2)×1 is the tracking error of the N-joint snake-like robot; The second-order differential of the tracking error of the N-joint snake-like robot is: in, is the second-order differential of e(t), yes The second-order differential of Yes φ The second-order differential of (t); definition Matrix, the dynamic model of the N-joint snake-like robot is changed as follows: in, is the dimensionless parameter gain matrix, Substitute the second-order differential of the tracking error to construct the tracking error dynamics model of the N-joint snake-like robot:

3. The adaptive global fractional-order terminal sliding mode snake-like robot control method according to claim 1 or 2, characterized in that: The radial basis neural network observer of S2 is designed as follows: The radial basis neural network observer is defined as follows: in, is all the unknown and uncertain terms in the tracking error dynamics model of the N-joint snake-like robot F = W * The approximate value of h(x)+Γ, where h(x) is the hidden layer activation function of the neural network observer. is the optimal weight of the neural network from the hidden layer to the output layer, Γ is between F and The approximate error of It is W * An estimated value of The hidden layer activation function is chosen as the cubic B-spline basis function, which is defined as: h(x)=[h1(t),h2(t),...,h N (t),h N+1 (t),h N+2 (t)] When || xo i ||≤h i When i=1, 2…, N+2 When h i <||xo i ||≤2h i When i=1, 2…, N+2 When 2h i <||xo i ||, and i=1, 2…, N+2 h i (x)=0; Among them, ||xo i || is the distance between the input vector and the center vector in Euclidean form; x is the input vector; o and o i is the center vector; h i is the width of the cubic B-spline function.

4. The adaptive global fractional-order terminal sliding mode snake-like robot control method according to claim 1 or 2, characterized in that: The global fractional-order terminal sliding surface of S3 is designed as follows: The global fractional-order terminal sliding surface is designed as: s(t)=[s1(t),s2(t),...,s N (t),s N+1 (t),s N+2 (t)] T n = diag(n1, n2,..., n N ,or N+1 ,or N+2 ) γ=diag(γ1,γ2,...,γ N ,c N+1 ,Y N+2 ) e(t)=[e1(t),e2(t),...,e N (t),e N+1 (t),e N+2 (t)] T e(0)=[e1(0),e2(0),...,e N (0),e N+1 (0),e N+2 (0)] T Among them, ηe(t) is the tracking error proportional term of the N-joint snake-like robot, and η is the proportional term parameter adjustment gain; is the fractional-order integral term, Υ is the fractional-order term parameter gain, D ε is the fractional integral, ε is the fractional integral order; The first-order differential of the global fractional-order terminal sliding surface is as follows: in, is the first-order differential of s(t) with respect to time t, They are s1(t),s2(t),…,s N (t),s N+1 (t),s N+2 (t)The first-order differential with respect to time t.

5. The adaptive global fractional-order terminal sliding mode snake-like robot control method according to claim 1 or 2, characterized in that: The constant velocity approach rate of S4 is designed as follows: The adaptive constant speed approach rate is designed as follows: in, is the adaptive term parameter of the constant velocity approach rate, The adaptive rate is designed as: Where ζ is the adaptive rate parameter gain, ζ=diag(ζ1,ζ2,…,ζ N ,ζ N+1 ,ζ N+2 ).

6. The adaptive global fractional-order terminal sliding mode snake-like robot control method according to claim 1 or 2, characterized in that: The controller design of the S5 is as follows: Combining the constructed radial basis neural network observer, the global fractional-order terminal sliding mode surface and the adaptive constant velocity approach rate, the adaptive global fractional-order terminal sliding mode controller of the N-joint snake robot system based on the neural network observer is designed as follows:

Images