Robot anti-interference control method based on neural network interference observer

Abstract

The invention discloses a robot anti-interference control method based on a neural network interference observer, and the method specifically comprises the following steps: 1, building a robot dynamics model containing external interference, and constructing a state space model; step 2, establishing a fixed time interference observer based on a radial basis function neural network, and performing external interference estimation on the robot kinetic model; and step 3, a fixed time sliding mode controller is designed based on the interference estimation result, and fixed time control under the condition of external interference is realized. According to the robot anti-interference control method based on the neural network interference observer, the interference is accurately and rapidly estimated by using the fixed time interference observer, the convergence of the system state in the fixed time is realized by combining the fixed time sliding mode controller, and the feedforward compensation and feedback suppression of the interference are combined, so that the system stability is improved. The method has the advantages of simple implementation mode, high interference estimation precision and high response speed.

Description

Technical Field

[0001] This invention relates to the field of robot control technology, specifically to a robot anti-interference control method based on a neural network interference observer. Background Technology

[0002] Robots, as crucial remote control and automation tools in modern society, have been widely applied in various fields such as industrial manufacturing, medical surgery, nuclear energy maintenance, disaster relief, and aerospace exploration. Compared to manual operation, robots can not only replace humans in completing high-risk tasks in dangerous or complex environments but also improve operational efficiency and execution accuracy. Therefore, their control performance directly affects the reliability and safety of task completion. However, in actual operation, robots are often affected by multiple internal and external disturbances, such as joint friction, gear backlash, load changes, sensor noise, and external environmental disturbances. These disturbances are transmitted to the system body through robot dynamics coupling, easily causing joint motion deviations, resulting in inaccurate positioning of the end effector, and reducing the overall control performance of the robot. Due to their powerful nonlinear approximation capabilities, neural networks have been introduced into the field of disturbance modeling and observer design in recent years, enabling high-precision disturbance estimation even when the disturbance mechanism is unknown or the data is incomplete. However, most existing methods rely on asymptotic convergence characteristics and cannot achieve fast and effective disturbance compensation within a limited time, making it difficult to fully guarantee the dynamic performance and stability of robots in complex environments. Therefore, how to design a fixed-time anti-interference control method based on a neural network interference observer to achieve rapid estimation and accurate compensation of multi-source interference has become a key technical problem that urgently needs to be solved in the field of robot control.

[0003] Currently, research on fixed-time control of robots based on neural network observers remains limited. Chinese patent application 202110413699.4 proposes an event-triggered adaptive fixed-time control method for a robotic arm, which can achieve system stability within a fixed time and, to some extent, reduce resource waste and improve tracking speed and accuracy during control. However, this method fails to effectively introduce an interference observer, resulting in insufficient handling of interference and thus limiting its ability to further improve control accuracy. Chinese patent application 202510445207.8 discloses a control method and system for a dual-link robotic arm based on a neural network observer. It utilizes the learning ability of reinforcement learning and combines gradient descent to design the adaptive update rate of the Actor-Critic neural network, enabling the approximate controller to approximate the actual controller. However, this scheme is not combined with a fixed-time control method, making it difficult to achieve rapid control.

[0004] Based on the search of the above materials, it can be seen that existing methods either lack the effective introduction of interference observers, resulting in limited anti-interference capabilities, or fail to integrate with fixed-time control strategies, thus failing to achieve fast and accurate control of the robot system. In view of this, a robot anti-interference control method based on neural network interference observers is proposed. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a robot anti-interference control method based on a neural network interference observer, which solves the problems mentioned in the background section.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a robot anti-interference control method based on a neural network interference observer, specifically comprising the following steps:

[0007] Step 1: Establish a robot dynamics model that includes external disturbances and construct a state-space model;

[0008] Step 2: Establish a fixed-time disturbance observer based on radial basis function neural network to estimate external disturbances in the robot dynamics model;

[0009] Step 3: Design a fixed-time sliding mode controller based on the disturbance estimation results to achieve fixed-time control under external disturbance conditions.

[0010] The present invention is further configured such that the expression of the robot dynamics model is:

[0011]

[0012] in, It is the joint position vector. and They are The first and second time derivatives, It is the robot's inertia matrix. It is the matrix of Coriolis force and centripetal force. The gravity vector It is the torque vector of friction and other disturbing forces. To control the torque.

[0013] The present invention is further configured such that the expression of the state-space model is:

[0014]

[0015] in , , and State variables and The first derivative with respect to time This represents the total disturbance torque vector in the state-space model. This represents the robot's inertia matrix in the state-space model. express The inverse matrix, This represents the coefficient matrix of the Coriolis force and centripetal force in the state-space model. To control the torque.

[0016] The present invention is further configured such that the formula for calculating the disturbance torque vector is:

[0017]

[0018] In the formula, This represents the total disturbance torque vector in the state-space model. This represents the robot's inertia matrix in the state-space model. express The inverse matrix, This represents the gravity vector in the state-space model. This is the torque vector of friction and other disturbance forces in the state-space model.

[0019] The present invention is further configured such that the expression for the fixed-time interference observer is:

[0020]

[0021] In the formula, , , and State variables and The first derivative with respect to time Yes The estimate, This represents the total disturbance torque vector in the state-space model. Yes The estimate, Let be the radial basis vector weight matrix. Yes The estimate, Yes The first time derivative, This represents the robot's inertia matrix in the state-space model. express The inverse matrix, This represents the coefficient matrix of the Coriolis force and centripetal force in the state-space model. To control the torque, , , , and They are positive numbers. It is a state The estimated value The estimation error is expressed as , , and These are the error vectors. The first, second, and third dimension elements, It is represented as, , Represented as The superscript "T" indicates the transpose operation, and the superscript "T" indicates the transpose operation. "and" "All are exponents and satisfy respectively" and , For absolute values, For symbolic functions, , ,in, , and They are vectors The first, second, and third elements, with the superscript "T" indicating the transpose operation.

[0022] The present invention is further configured such that the expression for the fixed-time sliding mode controller is:

[0023]

[0024]

[0025]

[0026] In the formula, For sliding surface, and All are positive numbers. , and They are state vectors The first, second, and third elements, with the superscript "T" indicating the transpose operation. and All are exponents, and , , For absolute values, It is a symbolic function.

[0027] The present invention is further configured such that the expression of the fixed-time anti-interference controller is:

[0028]

[0029] In the formula, and These are all positive numbers.

[0030] This invention provides a robot anti-interference control method based on a neural network interference observer. It has the following beneficial effects:

[0031] This invention achieves accurate and rapid estimation of disturbances by utilizing a fixed-time disturbance observer and combines it with a fixed-time sliding mode controller to achieve system state convergence within a fixed time. It combines disturbance feedforward compensation and feedback suppression, and has the advantages of simple implementation, high disturbance estimation accuracy, and fast response speed, making it suitable for robot motion control applications. Attached Figure Description

[0032] Figure 1 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0033] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0034] Please see Figure 1 This invention provides a robot anti-interference control method based on a neural network interference observer, specifically including the following steps:

[0035] Step 1: Establish a robot dynamics model incorporating external disturbances. Consider a planar three-bar linkage robot system and the environmental disturbances it faces, such as friction, external impacts, and vibrations. The expression for the robot dynamics model is as follows:

[0036]

[0037] in, It is the joint position vector. and They are The first and second time derivatives, It is the robot's inertia matrix. It is the matrix of Coriolis force and centripetal force. The gravity vector It is the torque vector of friction and other disturbing forces. To control the torque;

[0038] A state-space model is constructed based on the robot dynamics model, and state variables are defined. , The above robot dynamics model is transformed into:

[0039]

[0040] In the formula, for The first time derivative, i.e. Further derivation yields:

[0041]

[0042] In the formula, express The inverse matrix, This represents the robot's inertia matrix in the state-space model. This represents the coefficient matrix of the Coriolis force and centripetal force in the state-space model. and State variables and The first derivative with respect to time Let be the disturbance torque vector in the state-space model. The expression for the state-space model is as follows:

[0043]

[0044] To further explain, the formula for calculating the disturbance torque vector is:

[0045]

[0046] In the formula, This represents the total disturbance torque vector in the state-space model. This represents the robot's inertia matrix in the state-space model. express The inverse matrix, This represents the gravity vector in the state-space model. This is the torque vector of friction and other disturbance forces in the state-space model.

[0047] Step 2: Establish a fixed-time disturbance observer based on a radial basis function neural network, where the radial basis function neural network possesses... The _ neuron, the _ ... The expression for the radial basis function kernel of a neuron is:

[0048]

[0049] In the formula, For the input vector, For the first The center vector of each neuron For the first The width of a neuron, It is the natural exponential function, and the superscript "T" indicates the transpose operation.

[0050] Radial basis function neural networks (RBNs) are based on a three-layer structure. The radial basis functions in the hidden layers possess local response characteristics, enabling them to accurately capture local fluctuations in interference. The output layer effectively integrates the nonlinear characteristics of the interference by linearly weighting and combining these local responses, ultimately achieving accurate fitting of complex interference signals. Therefore, the estimated interference moment vector... It can be written as:

[0051]

[0052] In the formula, The radial basis vector weight matrix; To estimate the residuals;

[0053] To achieve accurate estimation of disturbances, the radial basis vector weight matrix needs to be estimated. Therefore, by assembling the corresponding state-space model, a fixed-time disturbance observer based on a radial basis neural network is obtained. The expression for the fixed-time disturbance observer is as follows:

[0054]

[0055] In the formula, , , and State variables and The first derivative with respect to time Yes The estimate, This represents the total disturbance torque vector in the state-space model. Yes The estimate, Let be the radial basis vector weight matrix. Yes The estimate, Yes The first time derivative, This represents the robot's inertia matrix in the state-space model. express The inverse matrix, This represents the coefficient matrix of the Coriolis force and centripetal force in the state-space model. To control the torque, , , , and They are positive numbers. It is a state The estimated value The estimation error is expressed as , , and These are the error vectors. The first, second, and third dimension elements, It is represented as, , Represented as The superscript "T" indicates the transpose operation, and the superscript "T" indicates the transpose operation. "and" "All are exponents and satisfy respectively" and , For absolute values, For symbolic functions, , ,in, , and They are vectors The first, second, and third elements, with the superscript "T" indicating the transpose operation.

[0056] External disturbances are estimated for the robot dynamics model using a fixed-time disturbance observer.

[0057] Step 3: Design a fixed-time sliding mode controller based on disturbance estimation, with the following expression:

[0058]

[0059]

[0060]

[0061] In the formula, For sliding surface, and All are positive numbers. , and They are state vectors The first, second, and third elements, with the superscript "T" indicating the transpose operation. and All are exponents, and , , For absolute values, It is a symbolic function;

[0062] Based on the above sliding surface design, a fixed-time anti-interference controller is defined as follows:

[0063]

[0064] In the formula, and They are all positive numbers;

[0065] Fixed-time anti-interference controller is used to achieve fixed-time control under external interference.

[0066] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A robot anti-interference control method based on a neural network interference observer, characterized in that, Specifically, the following steps are included: Step 1: Establish a robot dynamics model that includes external disturbances and construct a state-space model; Step 2: Establish a fixed-time disturbance observer based on radial basis function neural network to estimate external disturbances in the robot dynamics model; Step 3: Design a fixed-time sliding mode controller based on the disturbance estimation results to achieve fixed-time control under external disturbance conditions.

2. The robot anti-interference control method based on a neural network interference observer according to claim 1, characterized in that, The expression for the robot dynamics model is: ； in, It is the joint position vector. and They are The first and second time derivatives, It is the robot's inertia matrix. It is the matrix of Coriolis force and centripetal force. The gravity vector It is the torque vector of friction and other disturbing forces. To control the torque.

3. The robot anti-interference control method based on a neural network interference observer according to claim 2, characterized in that, The expression for the state-space model is: ； in , , and State variables and The first derivative with respect to time This represents the total disturbance torque vector in the state-space model. This represents the robot's inertia matrix in the state-space model. express The inverse matrix, This represents the coefficient matrix of the Coriolis force and centripetal force in the state-space model. To control the torque.

4. The robot anti-interference control method based on a neural network interference observer according to claim 3, characterized in that, The formula for calculating the disturbance torque vector is: ； In the formula, This represents the total disturbance torque vector in the state-space model. This represents the robot's inertia matrix in the state-space model. express The inverse matrix, This represents the gravity vector in the state-space model. This is the torque vector of friction and other disturbance forces in the state-space model.

5. The robot anti-interference control method based on a neural network interference observer according to claim 4, characterized in that, The expression for the fixed-time interference observer is: ； In the formula, , , and State variables and The first derivative with respect to time Yes The estimate, This represents the total disturbance torque vector in the state-space model. Yes The estimate, Let be the radial basis vector weight matrix. Yes The estimate, Yes The first time derivative, This represents the robot's inertia matrix in the state-space model. express The inverse matrix, This represents the coefficient matrix of the Coriolis force and centripetal force in the state-space model. To control the torque, , , , and They are positive numbers. It is a state The estimated value The estimation error is expressed as , , and These are the error vectors. The first, second, and third dimension elements, It is represented as, , Represented as The superscript "T" indicates the transpose operation, and the superscript "T" indicates the transpose operation. "and" "All are exponents and satisfy respectively" and , For absolute values, For symbolic functions, , ,in, , and They are vectors The first, second, and third elements, with the superscript "T" indicating the transpose operation.

6. The robot anti-interference control method based on a neural network interference observer according to claim 5, characterized in that, The radial basis neural network has The _ neuron, the _ ... The expression for the radial basis function kernel of a neuron is: ； In the formula, For the input vector, For the first The center vector of each neuron For the first The width of a neuron, It is the natural exponential function, and the superscript "T" indicates the transpose operation.

7. The robot anti-interference control method based on a neural network interference observer according to claim 6, characterized in that, The expression for the fixed-time sliding mode controller is: ； ； ； In the formula, For sliding surface, and All are positive numbers. , and They are state vectors The first, second, and third elements, with the superscript "T" indicating the transpose operation. and All are exponents, and , , For absolute values, It is a symbolic function.

8. The robot anti-interference control method based on a neural network interference observer according to claim 7, characterized in that, The expression for the fixed-time anti-interference controller is: ； In the formula, and These are all positive numbers.

Images