Control method and application of flexible joint collaborative robot based on adaptive jump

Abstract

The present invention discloses a control method and application of a flexible joint collaborative robot based on adaptive jump. The control method comprises: calculating a combined error; calculating a feedforward compensation term and a feedback control term; adding the feedforward and feedback terms for control; wherein the combined error comprises first-order to third-order errors; the adaptive jump control comprises a PID sub-item and a RISE sub-item, the RISE sub-item comprises an adaptive sliding mode gain, and the adaptive sliding mode gain is related to the second-order error and the first-order differential of the second-order error. The method provided by the present invention reduces the computational complexity, improves the real-time computing capability of the system, and enables the control system to maintain a high response speed under high dynamic conditions; the design of the adaptive jump improves the system's resistance to uncertainty and external disturbances, and significantly reduces the high-frequency flutter problem caused by the flexible joints; simplifies the adjustment process of the control parameters, and is particularly suitable for the rapid deployment of multi-degree-of-freedom collaborative robots.

Description

Technical Field

[0001] The present invention relates to the field of robotics technology, and in particular to a control method and application of a flexible joint collaborative robot based on adaptive jump. Background Art

[0002] Compared with the bulky design and closed application of traditional industrial robots, collaborative robots are open-chain mechanisms composed of multiple integrated flexible joints and lightweight connecting rods in series, which can improve the safety and efficiency of collaborative operations between robots and people in open environments. They have been widely used in intelligent manufacturing, medical and education industries, but the lower system stiffness caused by flexible joints will affect the robot tracking error. Without changing the structure, the tracking accuracy can be improved through motion control technology, an economical way. However, there are several significant problems with the tracking control of flexible joint collaborative robots:

[0003] (1) Most flexible joint robot controllers use decentralized control (such as Chinese invention patents with publication numbers CN113183154A, CN112643670A, and CN108181813A, etc.), that is, a full-order dynamic model of a single flexible joint is established, and based on this model, a controller such as a backstepping-based adaptive controller, a sliding mode observer, and a fractional-order sliding mode controller is designed to compensate for the flexible modal interference of the joint and achieve stable signal tracking control of the joint. However, this method ignores the dynamic coupling characteristics between the various link joint modules, resulting in low overall positioning accuracy of the robot. On the other hand, pure decentralized control requires multiple control parameter settings for each joint, which is not conducive to the rapid deployment of multi-degree-of-freedom collaborative robots.

[0004] (2) Establish a full-order dynamic model of the flexible joint robot. This method can reveal the force-position response characteristics between each joint module and the flexible mode inside the joint. Based on this, the control law of the system is designed by using virtual control based on backstepping (such as the Chinese invention patent with publication number CN114839880A), dual-time scale system decoupling based on singular perturbation (such as the Chinese invention patent with publication number CN112720456A), and motor and connecting rod end system decoupling (such as the Chinese invention patent with publication number CN106335064A). However, these methods use relatively complex high-order models, and their real-time computing efficiency is low, making it difficult to meet application requirements.

[0005] (3) The reduced-order rigid body dynamics model has a low complexity and can reflect the coupling dynamics between joint modules. Based on this, the joint flexible mode is regarded as an external disturbance, which generally needs to be compensated by adaptive neural network estimator (such as Chinese invention patent with publication number CN110877333A), sliding mode controller (such as Chinese invention patent with publication number CN117908362A) and other methods. However, factors such as slow convergence of the estimation error of the estimator and rapid switching of the sliding mode control law causing high-frequency vibration of the joints can easily lead to low control accuracy. Summary of the invention

[0006] In view of the deficiencies in the prior art, the object of the present invention is to provide a flexible joint collaborative robot control method and application based on adaptive jump.

[0007] In order to achieve the above-mentioned invention object, the technical solution adopted by the present invention includes:

[0008] In a first aspect, the present invention provides a control method for a flexible joint collaborative robot based on adaptive jump, which comprises:

[0009] Based on the expected tracking trajectory and actual operation point displacement of the flexible joint collaborative robot, an error filter is used to calculate the combined error;

[0010] Based on the combined error, a radial basis neural network model is used to calculate a feedforward compensation term, and an adaptive jump control is used to calculate a feedback control term;

[0011] Adding the feedforward compensation term and the feedback control term to control the drive motor of the flexible joint collaborative robot;

[0012] The combined error includes a first-order error, a second-order error and a third-order error, the first-order error is a tracking error between the desired tracking trajectory and the actual operating point displacement, the second-order error is linearly related to the first-order error and the first-order differential of the first-order error, and the third-order error is linearly related to the second-order error and the first-order differential of the second-order error;

[0013] The adaptive jump control includes a PID sub-item and a RISE sub-item. The RISE sub-item adopts robust integral sign control and includes an adaptive sliding mode gain and a constant sliding mode gain. The adaptive sliding mode gain has a correlation with the second-order error and the first-order differential of the second-order error.

[0014] In a second aspect, the present invention further provides a design method for the above-mentioned flexible joint collaborative robot control method, which comprises:

[0015] (1) Based on the reduced-order rigid body dynamics model, the error dynamics model is obtained;

[0016] (2) defining the tracking error of the flexible joint collaborative robot in the operating space, and designing a combined error filter based on velocity and acceleration signals;

[0017] (3) Based on the combined error and radial basis neural network model, an adaptive feedforward compensation controller is designed; a robust integral symbol controller is designed using adaptive jumps; the feedforward compensation controller and the robust integral symbol controller are combined to form a binary control architecture of "adaptive feedforward + adaptive jump feedback";

[0018] (4) Analyze the combined error filter, select state quantities, design a closed-loop error state equation, construct a Lyapunov function from the closed-loop error state equation, and prove the asymptotic convergence of the tracking error of the closed-loop system based on the Lyapunov stability theory to verify the control method of the flexible joint collaborative robot.

[0019] In a third aspect, the present invention further provides a flexible joint collaborative robot control system based on adaptive jump, which comprises:

[0020] A combined error module is used to calculate the combined error based on the expected tracking trajectory and actual operation point displacement of the flexible joint collaborative robot using an error filter;

[0021] A control item calculation module, comprising a feedforward compensation unit and a feedback control unit; based on the combined error, the feedforward compensation unit is used to calculate the feedforward compensation item using a radial basis neural network model, and the feedback control unit is used to calculate the feedback control item using adaptive jump control;

[0022] A binary control module, used for adding the feedforward compensation term and the feedback control term to control the drive motor of the flexible joint collaborative robot;

[0023] The combined error module includes a first-order error unit, a second-order error unit and a third-order error unit. The first-order error unit is used to obtain the tracking error between the desired tracking trajectory and the actual operation point displacement. The second-order error unit is used to linearly add the first-order error and the first-order differential of the first-order error to obtain the second-order error. The third-order error unit is used to linearly add the second-order error and the first-order differential of the second-order error to obtain the third-order error.

[0024] The feedback control unit includes a PID subunit and a RISE subunit. The RISE subunit adopts a robust integral sign control logic and includes an adaptive sliding mode gainer and a constant sliding mode gainer. The adaptive sliding mode gainer is used to calculate an adaptive sliding mode gain that is correlated with the second-order error and the first-order differential of the second-order error.

[0025] Based on the above technical solution, compared with the prior art, the beneficial effects of the present invention at least include:

[0026] The flexible joint collaborative robot control method provided by the present invention reduces the computational complexity, improves the real-time computing capability of the system, and enables the control system to maintain a high response speed under high dynamic conditions; the adaptive jump design improves the system's resistance to uncertainty and external disturbances, and significantly reduces the high-frequency vibration problem caused by flexible joints; and simplifies the adjustment process of control parameters, which is particularly suitable for the rapid deployment of multi-degree-of-freedom collaborative robots.

[0027] The above description is only an overview of the technical solution of the present invention. In order to enable those skilled in the art to more clearly understand the technical means of the present application and implement them according to the contents of the specification, the following is a description of the preferred embodiments of the present invention in conjunction with detailed drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a functional composition diagram of a flexible joint collaborative robot control method based on adaptive jump provided by a typical embodiment of the present invention;

[0029] Figure 2 It is a "feedforward + feedback" control logic block diagram of a flexible joint collaborative robot control method based on adaptive jump provided by a typical embodiment of the present invention;

[0030] Figure 3 It is the hardware composition architecture of a flexible joint collaborative robot control method system based on adaptive jump provided by a typical embodiment of the present invention. DETAILED DESCRIPTION

[0031] In view of the deficiencies in the prior art, the inventor of this case has proposed the technical solution of the present invention after long-term research and extensive practice. The technical solution, its implementation process and principle will be further explained as follows.

[0032] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0033] Moreover, relational terms such as “first” and “second” are merely used to distinguish one component or method step from another with the same name, but do not necessarily require or imply any such actual relationship or order between these components or method steps.

[0034] like Figure 1 and Figure 2As shown, an embodiment of the present invention provides a flexible joint collaborative robot control method based on adaptive jump, which includes the following steps:

[0035] Based on the expected tracking trajectory and actual operation point displacement of the flexible joint collaborative robot, an error filter is used to calculate the combined error;

[0036] Based on the combined error, a radial basis neural network model is used to calculate a feedforward compensation term, and an adaptive jump control is used to calculate a feedback control term;

[0037] Adding the feedforward compensation term and the feedback control term to control the drive motor of the flexible joint collaborative robot;

[0038] The combined error includes a first-order error, a second-order error and a third-order error, the first-order error is a tracking error between the desired tracking trajectory and the actual operating point displacement, the second-order error is linearly related to the first-order error and the first-order differential of the first-order error, and the third-order error is linearly related to the second-order error and the first-order differential of the second-order error;

[0039] The adaptive jump control includes a PID sub-item and a RISE sub-item. The RISE sub-item adopts robust integral sign control and includes an adaptive sliding mode gain and a constant sliding mode gain. The adaptive sliding mode gain has a correlation with the second-order error and the first-order differential of the second-order error.

[0040] The main improvement of the present invention is reflected in the improvement of the adaptive jump control in the RISE sub-item. In this regard, in some specific implementations, the calculation method of the adaptive sliding mode gain can be expressed as:

[0041] ;

[0042] in, represents the adaptive sliding mode gain; represents the second-order error; is a positive constant, is a symbolic vector function; express The transposed form of the vector first-order differential of ; represents the first-order differential; and is the coefficient.

[0043] Based on the above improvements, in some implementations, the differential expression of the adaptive jump control is:

[0044] ;

[0045] in, represents the feedback control item; represents the proportional control gain matrix, is a natural number; Represents the identity matrix, dimension is n ; represents the third-order error; represents the constant sliding mode gain.

[0046] Regarding other aspects that cooperate with the above-mentioned main improvements, in some embodiments, the calculation method of the combined error is expressed as:

[0047] ;

[0048] in, represents the tracking error; represents the expected tracking trajectory; represents the actual operating point displacement; represents the first-order error; represents the second-order error; represents the third-order error; represents the combined error; is the coefficient.

[0049] In some implementations, the calculation of the feedforward compensation term is expressed as:

[0050] ;

[0051] in, represents the feedforward compensation term; represents the radial basis network activation function matrix, T Represents matrix transpose; represents the network parameter vector, Indicates the number of nodes in the output layer; is the number of nodes representing the hidden layer; represents a constant gain parameter; Represents the network parameter vector updated in real time;

[0052] The radial basis network activation matrix and network parameter vector are expressed as:

[0053] ;

[0054] in, represents the kernel function of the hidden layer; represents the identity matrix; represents the ideal network connection weight; The vector representing the network input layer, i.e., the vector consisting of joint angles and velocities; represents the generalized coordinates of the flexible joint collaborative robot; represents the Kronecker product operation; Represents vectorized operations on matrices.

[0055] Under the specific feedback+feedforward control logic, in some implementations, the drive motor may be controlled by at least a binary control input, and the calculation method of the binary control input is expressed as:

[0056]

[0057] in, represents the binary control input.

[0058] In addition, regarding the theoretical model based on the overall design, in some embodiments, the dynamic model based on the flexible joint collaborative robot control method only includes a reduced-order rigid body dynamics model, and the reduced-order rigid body dynamics model is expressed as:

[0059] ;

[0060] in, represents the generalized coordinates of the flexible joint collaborative robot; represents the inertia matrix; represents the Coriolis-centrifugal force term; represents the gravity vector; represents the control input of the operation space, Represents the result of pseudo-inverse transformation and transposition of Jacobian matrix J(q); represents the lumped disturbance, which includes the uncertainty, unmodeled flexible dynamics and external disturbance of the flexible joint collaborative robot.

[0061] Further in some embodiments, the closed-loop error state equation of the flexible joint collaborative robot is expressed as:

[0062] ;

[0063] in, Represents the network parameter estimation error; Represents the uncertainty term.

[0064] Correspondingly, the second aspect of the embodiment of the present invention further provides a design method for the flexible joint collaborative robot control method provided in any of the above embodiments, which comprises the following steps:

[0065] (1) Based on the reduced-order rigid body dynamics model, the error dynamics model is obtained;

[0066] (2) defining the tracking error of the flexible joint collaborative robot in the operating space, and designing a combined error filter based on velocity and acceleration signals;

[0067] (3) Based on the combined error and radial basis neural network model, an adaptive feedforward compensation controller is designed; a robust integral symbol controller is designed using adaptive jumps; the feedforward compensation controller and the robust integral symbol controller are combined to form a binary control architecture of "adaptive feedforward + adaptive jump feedback";

[0068] (4) Analyze the combined error filter, select state quantities, design a closed-loop error state equation, construct a Lyapunov function from the closed-loop error state equation, and prove the asymptotic convergence of the tracking error of the closed-loop system based on the Lyapunov stability theory to verify the control method of the flexible joint collaborative robot.

[0069] The design method and the control method of the flexible joint collaborative robot provided by the present invention simplify the complexity of the system through a reduced-order rigid body dynamics model, and perform control design in combination with an error dynamics model to ensure that the control system has fast response capability and strong robustness. The method defines the tracking error of the robot in the operating space and designs an error filter to reduce the influence of the flexible joint modal excitation. In the design of the control system, the present invention adopts a radial basis neural network (RBFNN) model, realizes feedforward control compensation by online learning of the robot dynamics model, and designs an adaptive feedforward controller based on this to compensate for the uncertainty of the system. At the same time, the robust feedback control further enhances the stability of the system in an uncertain environment through adaptive jump design. By combining adaptive feedforward and robust feedback control, the present invention forms an effective binary control architecture to ensure the high-precision control performance of the system in a complex operating environment. In order to verify the stability of the system, the present invention constructs a stability proof of a closed-loop system based on the Lyapunov function, indicating that the tracking error can converge asymptotically over time, thereby ensuring the robustness and tracking accuracy of the control system. In addition, the method enables the robot to effectively resist external disturbances in an uncertain environment and achieve smooth tracking of the desired trajectory through error compensation design.

[0070] Furthermore, a third aspect of the embodiment of the present invention further provides a flexible joint collaborative robot control system based on adaptive jump, which includes:

[0071] A combined error module is used to calculate the combined error based on the expected tracking trajectory and actual operation point displacement of the flexible joint collaborative robot using an error filter;

[0072] A control item calculation module, comprising a feedforward compensation unit and a feedback control unit; based on the combined error, the feedforward compensation unit is used to calculate the feedforward compensation item using a radial basis neural network model, and the feedback control unit is used to calculate the feedback control item using adaptive jump control;

[0073] A binary control module, used for adding the feedforward compensation term and the feedback control term to control the drive motor of the flexible joint collaborative robot;

[0074] The combined error module includes a first-order error unit, a second-order error unit and a third-order error unit. The first-order error unit is used to obtain the tracking error between the desired tracking trajectory and the actual operation point displacement. The second-order error unit is used to linearly add the first-order error and the first-order differential of the first-order error to obtain the second-order error. The third-order error unit is used to linearly add the second-order error and the first-order differential of the second-order error to obtain the third-order error.

[0075] The feedback control unit includes a PID subunit and a RISE subunit. The RISE subunit adopts a robust integral sign control logic and includes an adaptive sliding mode gainer and a constant sliding mode gainer. The adaptive sliding mode gainer is used to calculate an adaptive sliding mode gain that is correlated with the second-order error and the first-order differential of the second-order error.

[0076] The technical solution of the present invention is further described in detail below through several embodiments and in conjunction with the accompanying drawings. However, the selected embodiments are only used to illustrate the present invention, and do not limit the scope of the present invention. It should be noted that some constant parameters in the present invention are not important features of the technical solution provided by the present invention. Those skilled in the art are fully capable of obtaining the corresponding optimal parameter combination through a limited number of experiments based on the technical concept provided by the present invention, and are not limited to the scope of the examples in the present invention.

[0077] Example 1

[0078] This embodiment illustrates the design process of the flexible joint collaborative robot control method described above, which is specifically as follows:

[0079] Step (1)

[0080] Considering the reduced-order rigid body dynamics model of the collaborative robot in the joint space, the expression is:

[0081] ;

[0082] in, are the generalized coordinates of the robot; , and They are the inertia matrix, the Coriolis-centrifugal term, and the gravity vector; is the control input of the joint; It is defined as the lumped disturbance consisting of the robot’s model uncertainty, unmodeled flexible dynamics, and external disturbances.

[0083] The differential motion equation of the robot is expressed as:

[0084] ;

[0085] in, is the actual operation point displacement in the operation space; is the Jacobian matrix of the robot, m is the number of joints, and n is the dimension of the operation space. · represents the first-order differential, ·· represents the second-order differential.

[0086] Substituting equation (2) into equation (1), we can obtain the dynamic equation of the robot in the operating space, which is expressed as:

[0087] ;

[0088] Among them, in the equation , ,as well as ; and ; is the generalized inverse of the matrix. for The transposed matrix of .

[0089] Step (2)

[0090] Define the tracking error of the robot in the operating space as:

[0091] ;

[0092] in is the desired tracking trajectory, is the tracking error, is the speed error, is the acceleration error.

[0093] In order to avoid exciting the soft mode of the system, an error filter is defined, and its expression is:

[0094] ;

[0095] in, is the combination of displacement and velocity error, is the combination of velocity and acceleration, is a positive filter constant. Solving the differential equation of formula (5), we can get

[0096] ;

[0097] in and is the initial value. Obviously, if , and will converge to the initial value at an exponential rate. The control objective is to make Converges to zero, allowing the robot to smoothly track the reference trajectory.

[0098] Define a variable that represents the lumped combined error:

[0099] (7)

[0100] Step (3)

[0101] Set up the following binary control inputs:

[0102] (8)

[0103] in, is the feedback control item; is the feed-forward compensation term.

[0104] First, based on the feedforward control algorithm of adaptive RBFNN, the feedforward compensation term is designed. Through online learning, the dynamic model is obtained, and the expression is as follows:

[0105] (9)

[0106] The radial basis neural network is introduced, and its network model is defined as follows:

[0107] (10)

[0108] in, is the ideal connection weight matrix, is the kernel function of the hidden layer, is the number of nodes in the output layer, is the number of nodes in the hidden layer. is the vector of the input layer. is the bounded bias of the neural network approximation. The Gaussian kernel function is selected as follows:

[0109] (11)

[0110] in and They are the center vector of the perceptual field of view and the hyperparameters of the constant value.

[0111] In order to update the parameters of the weight matrix , Rearranged into a linear parameterized form, as follows:

[0112] (12)

[0113] in , and Therefore, the feedforward compensation can be obtained as:

[0114] (13)

[0115] in is the weight parameter to be estimated, is the estimation law of weight parameters, is the constant gain parameter of the update law.

[0116] Then, based on the adaptive jump feedback control algorithm, the feedback control term is designed. .

[0117] Substituting equation (9) into equation (3), the actual dynamic model is:

[0118] (14)

[0119] in Represents the deviation between the robot's expected dynamics and actual dynamics, and its expression is:

[0120] ;

[0121] (15)

[0122] In order to stabilize the acceleration-related , Substituting formula (5) into formula (14), we get:

[0123] (16)

[0124] In order to avoid soft mode excitation, Components The robust term in must be continuous. In other words, the control input derivative Must be bounded. Therefore, taking the time derivative of equation (16), we obtain a first-order nonlinear system:

[0125] (17)

[0126] in,

[0127] ;

[0128] (18)

[0129] RISE control is introduced to guide the design of control law:

[0130] (19)

[0131] in, Contains accelerations that are difficult to measure. pass and To solve. Formula (19) can be written as:

[0132] ;

[0133] (20)

[0134] in, Control gain matrix. is the adaptive sliding mode gain. And is a constant sliding mode gain. yes hour and The initial value of . The first term is the traditional PID control, while the second term is the adaptive RISE control. In order to balance the control energy and vibration suppression ability, the sliding mode gain should be adapted according to the size of the uncertainty. Therefore, the specially designed Automatic adjustment is performed according to the following formula.

[0135] (twenty one)

[0136] In the formula, and is a constant. Obviously, yes and The output of the forced system as input. Therefore, Can be based on If the uncertainty disappears, can converge to .

[0137] Furthermore, the solution of the adaptive gain is:

[0138] (twenty two)

[0139] In the formula is the initial value.

[0140] Step (4)

[0141] The error filter defined in step (2) has the following properties:

[0142] (twenty three)

[0143] Define the domain as State variables for:

[0144] (twenty four)

[0145] Define the Lyapunov function as follows:

[0146] (25)

[0147] Obviously there is a constant , Make . For the above formula Find the time derivative and simplify to get:

[0148] ;

[0149] (26)

[0150] By the following inequality relationship:

[0151] (27)

[0152] Available Satisfies the following inequality:

[0153] (28)

[0154] Among them, , , ,and The value of is selected according to the following inequality

[0155] (29)

[0156] Available , which shows and Bounded. Further, is also bounded. Therefore, by Barbalat's theorem, we can conclude that , as well as .therefore, , In other words, after theoretical verification, the tracking error of the designed closed-loop system can converge asymptotically.

[0157] Example 2

[0158] This embodiment uses the control method obtained by the design method in the above embodiment 1 to build an adaptive jump control system for a flexible joint robot. Its architecture is as follows: Figure 3 As shown. The joint motor position signal is measured by a 19-bit encoder and fed back to the Speedgoat real-time controller. The reference trajectory and control algorithm are programmed in MATLAB / Simulink, and the host computer real-time interactive interface is designed based on this; thereby achieving real-time operation of the collaborative robot system. At the same time, the host computer exchanges information with the Speedgoat controller via Ethernet. The signal of the control algorithm is implemented by the Speedgoat hardware, and the desired current signal is sent to the driver. The driver has a current loop controller. The transmission ratios of the large and small joints are 160 and 100. The running frequency of the Simulink program is set to 1KHz. The speed and acceleration signals of each axis are obtained by passing the position signal through a first-order high-pass filter s / (0.004s+1).

[0159] Example 3

[0160] This embodiment uses the control method and system provided in Embodiments 1 and 2 to perform experimental tests on processing applications. The specific test items, test standards and test results are as follows:

[0161] In order to verify the tracking performance of the proposed control method and system, the test items are the expected trajectories in the horizontal and vertical directions of plane machining. m, m, ,The goal is to enable the flexible joint robot to track the desired trajectory. In order to quantify the test criteria, the ,maximum error, minimum error, root mean square error and other conventional indicators of the ,robot’s trajectory tracking are measured respectively. The specific test data are shown in Table 1.

[0162] Table 1. Minimum, maximum and root mean square error of trajectory tracking of Example 3 (unit: 10 -4 m)

[0163] .

[0164] Comparative Example 1

[0165] This comparative example is substantially the same as Examples 1-3, except that:

[0166] In formula (19), the adaptive sliding mode gain is removed .

[0167] The same application experiment as in Example 3 was used for testing, and the test data are shown in Table 2.

[0168] Table 2. Minimum, maximum and root mean square error of trajectory tracking for comparative example 1 (unit: 10 -4 m)

[0169] .

[0170] By comparing the results of Example 3 and Comparative Example 1, it is shown that the proposed control method and system have a smaller error and achieve a significant improvement effect.

[0171] Based on the above test results, it can be clearly seen that the flexible joint collaborative robot control method provided by the embodiment of the present invention reduces the computational complexity, improves the real-time computing capability of the system, and enables the control system to maintain a high response speed under high dynamic conditions; the design of adaptive jumps improves the system's resistance to uncertainty and external disturbances, and significantly reduces the high-frequency vibration problem caused by flexible joints; simplifies the adjustment process of control parameters, and is particularly suitable for the rapid deployment of multi-degree-of-freedom collaborative robots.

[0172] It should be understood that the above embodiments are only for illustrating the technical concept and features of the present invention, and their purpose is to enable people familiar with the technology to understand the content of the present invention and implement it accordingly, and they cannot be used to limit the protection scope of the present invention. Any equivalent changes or modifications made according to the spirit of the present invention should be included in the protection scope of the present invention.

Claims

1. A control method for a flexible joint collaborative robot based on adaptive jump, characterized in that: include: Based on the expected tracking trajectory and actual operation point displacement of the flexible joint collaborative robot, an error filter is used to calculate the combined error; Based on the combined error, the radial basis neural network model is used to calculate the feedforward compensation term, and the adaptive jump control is used to calculate the feedback control term; Adding feedforward compensation items and feedback control items to control the drive motor of the flexible joint collaborative robot; Among them, the combined error includes first-order error, second-order error and third-order error. The first-order error is the tracking error between the expected tracking trajectory and the actual operation point displacement. The second-order error is linearly related to the first-order error and the first-order differential of the first-order error. The third-order error is linearly related to the second-order error and the first-order differential of the second-order error. Adaptive jump control includes PID sub-items and RISE sub-items. The RISE sub-item adopts robust integral sign control, including adaptive sliding mode gain and constant sliding mode gain. The adaptive sliding mode gain is correlated with the second-order error and the first-order differential of the second-order error. The calculation method is expressed as: ; in, represents the adaptive sliding mode gain; represents the second-order error; is a positive constant, is a symbolic vector function; express The transposed form of the vector first-order differential of ; , is the coefficient; The differential expression of adaptive jump control is: ; in, represents the feedback control item; represents the proportional control gain matrix, is a natural number; Represents the identity matrix, dimension is n ; represents the third-order error; Represents the constant sliding mode gain.

2. The flexible joint collaborative robot control method according to claim 1, characterized in that: The calculation method of the combined error is expressed as: ; ; ; in, represents the tracking error; represents the expected tracking trajectory; represents the actual operating point displacement; represents the first-order error; represents the second-order error; represents the third-order error; represents the combined error; is the coefficient.

3. The flexible joint collaborative robot control method according to claim 1, characterized in that: The calculation of the feedforward compensation term is expressed as: ; in, represents the feedforward compensation term; represents the radial basis network activation function matrix, T Represents matrix transpose; represents the network parameter vector, Indicates the number of nodes in the output layer; is the number of nodes representing the hidden layer; represents a constant gain parameter; Represents the network parameter vector updated in real time; The radial basis network activation function matrix and network parameter vector are expressed as: ; ; ; in, represents the kernel function of the hidden layer; represents the identity matrix; represents the ideal network connection weight; The vector representing the network input layer, i.e., the vector consisting of joint angles and velocities; represents the generalized coordinates of the flexible joint collaborative robot; represents the Kronecker product operation; Represents vectorized operations on matrices.

4. The flexible joint collaborative robot control method according to claim 1, characterized in that: The drive motor is controlled by at least a binary control input, and the calculation method of the binary control input is expressed as: ; in, represents the binary control input.

5. The flexible joint collaborative robot control method according to any one of claims 1 to 4, characterized in that: The dynamic model based on which the flexible joint collaborative robot control method is described only includes a reduced-order rigid body dynamic model, and the reduced-order rigid body dynamic model is expressed as: ; ; in, Represents the inertia matrix of the operating space; represents the Coriolis-centrifugal force term in the operating space; Represents the gravity vector of the operating space; represents the control input of the operation space, Represents the result of pseudo-inverse transformation and transposition of Jacobian matrix J(q); represents the lumped disturbance, which includes the uncertainty, unmodeled flexible dynamics and external disturbance of the flexible joint collaborative robot.

6. The flexible joint collaborative robot control method according to claim 5, characterized in that: The closed-loop error state equation of the flexible joint collaborative robot is expressed as: ; in, Represents the network parameter estimation error; Represents the uncertainty term.

7. A design method for a flexible joint collaborative robot control method according to any one of claims 1 to 6, characterized in that: include: (1) Based on the reduced-order rigid body dynamics model, the error dynamics model is obtained; (2) defining the tracking error of the flexible joint collaborative robot in the operating space, and designing a combined error filter based on velocity and acceleration signals; (3) Based on the combined error and radial basis neural network model, an adaptive feedforward compensation controller is designed; a robust integral sign controller is designed using adaptive jumps; the feedforward compensation controller and the robust integral sign controller are combined to form a binary control architecture of adaptive feedforward + adaptive jump feedback; (4) Analyze the combined error filter, select state quantities, design a closed-loop error state equation, construct a Lyapunov function from the closed-loop error state equation, and prove the asymptotic convergence of the tracking error of the closed-loop system based on the Lyapunov stability theory to verify the control method of the flexible joint collaborative robot.

8. A flexible joint collaborative robot control system based on adaptive jump, used to implement the flexible joint collaborative robot control method according to any one of claims 1 to 6, characterized in that: include: A combined error module is used to calculate the combined error based on the expected tracking trajectory and actual operation point displacement of the flexible joint collaborative robot using an error filter; A control item calculation module, including a feedforward compensation unit and a feedback control unit; Based on the combined error, the feedforward compensation unit is used to calculate the feedforward compensation term by using a radial basis neural network model, and the feedback control unit is used to calculate the feedback control term by using adaptive jump control; A binary control module, used for adding the feedforward compensation term and the feedback control term to control the drive motor of the flexible joint collaborative robot; The combined error module includes a first-order error unit, a second-order error unit and a third-order error unit. The first-order error unit is used to obtain the tracking error between the desired tracking trajectory and the actual operation point displacement. The second-order error unit is used to linearly add the first-order error and the first-order differential of the first-order error to obtain the second-order error. The third-order error unit is used to linearly add the second-order error and the first-order differential of the second-order error to obtain the third-order error. The feedback control unit includes a PID subunit and a RISE subunit. The RISE subunit adopts a robust integral sign control logic and includes an adaptive sliding mode gainer and a constant sliding mode gainer. The adaptive sliding mode gainer is used to calculate an adaptive sliding mode gain that is correlated with the second-order error and the first-order differential of the second-order error.

Images