Industrial robot grinding and polishing compliance fuzzy control method and system based on weight-gravity / bit superposition control

Abstract

The invention discloses an industrial robot grinding and polishing compliance fuzzy control method based on weight-gravity / bit superposition control. The method comprises the following steps: establishing a robot Cartesian space dynamics model; establishing a selection matrix to decouple the Cartesian space; carrying out self-tuning on the core coefficient by utilizing a fuzzy control means; and obtaining control quantity to realize expected control of the robot. The invention provides a new method for robot polishing, and particularly for machining a workpiece with a complex profile, the method can realize self-tuning of core parameters and effectively solve the dynamic hysteresis in the polishing process, the expected force tracking capability and the expected position tracking capability of an end effector of an industrial robot are improved, and the method is suitable for industrial robot polishing. And the method has important significance in automatic machining and polishing operation effect control.

Description

Technical Field

[0001] The present invention belongs to the field of workpiece processing technology. More specifically, the present invention relates to a fuzzy control method and system for the grinding and polishing compliance of an industrial robot based on force / position control. Background Art

[0002] Modern high-end equipment manufacturing industry has increasingly higher requirements for the surface processing quality of workpieces. Industrial robot-based grinding has become a powerful means to achieve continuous contact operations such as grinding and polishing in high-end manufacturing. It is one of the important tools to solve the low efficiency of manual operations and the serious harm to the working environment, as well as to improve the quality of workpiece processing.

[0003] Robotic grinding operations require the robot's end-effector to contact the workpiece along a planned path. When grinding workpieces with complex curved features, excessive contact force between the robot's end-effector and the workpiece can damage the surface, while insufficient contact force can result in substandard machining. Therefore, precise control of compliance force is crucial for optimal results during robotic machining.

[0004] Based on the characteristics of robots achieving compliant motion, compliance force control can be classified into impedance control methods, force / position hybrid control methods, and intelligent control methods. Force / position hybrid control is also a passive compliance force control method. Compared with other methods, it can achieve a balance between force and position during task execution, adapt to complex dynamic environments, and improve operational stability and efficiency, thereby achieving better control of processing results. However, when dealing with curved workpieces with variable curvature, the current methods based on force / position hybrid control often have the problem of frequent switching of working modes during grinding and polishing due to the strong decoupling of the selection matrix. This can easily lead to problems such as slow response speed and dynamic hysteresis, affecting processing accuracy and work results.

[0005] Therefore, this embodiment proposes a fuzzy control method for the compliance force of industrial robot grinding and polishing based on weighted force / position superposition control, which effectively solves the hysteresis problem caused by strong decoupling of traditional force-position hybrid control methods under high dynamic working conditions and the problem of inability to simultaneously control force and position in one joint space. It provides new ideas for improving the efficiency and processing effect of industrial robot grinding and polishing, and is of great significance to the transformation and upgrading of the manufacturing industry, the automated production of high-precision equipment, and the intelligent control of equipment. Summary of the Invention

[0006] The present invention proposes a fuzzy control method and system for the grinding and polishing compliance of an industrial robot based on force / position control, aiming to solve the above problems.

[0007] First, this embodiment can realize the grinding of workpieces with complex curved surfaces. More specifically, the complex curved surfaces are workpieces with surface features having frequent curvature changes and high geometric freedom.

[0008] The present invention implements a method for fuzzy control of the compliance force of industrial robot grinding and polishing based on force / position control, and the method comprises the following steps:

[0009] Obtaining the desired force and the desired position as desired inputs of the method;

[0010] Establish a robot state prediction model, which includes the torque prediction equation, posture prediction equation and the Jacobian matrix of the robot body;

[0011] Establish a force control selection matrix to decouple the robot's joint space into force control subspace and position control subspace;

[0012] The robot's own state prediction model is used to convert the expected input into the robot's expected force control quantity, position control quantity and joint speed control quantity;

[0013] According to the selection matrix, the expected force control amount, position control amount and joint speed control amount of each joint are obtained;

[0014] The fuzzy controller is used to achieve self-tuning of the weight coefficient, thus improving adaptability and flexibility;

[0015] The corresponding expected control quantities are superimposed with weights to achieve simultaneous control of force and position, and obtain the control increment of the robot;

[0016] According to the control increment, the joint torque and joint angle of the robot are determined; according to the joint torque and joint angle, each joint of the robot is controlled and driven so that the robot can achieve the desired input.

[0017] In combination with the above method, an embodiment of the present invention provides an implementation method of the selection matrix mentioned in the above method, which includes: the Cartesian space where the robot end effector is located is a mixed dimension, which includes 6 degrees of freedom including the translation dimension and rotation dimension of three basic orthogonal coordinate axes.

[0018] More specifically, a 6-dimensional matrix is ​​selected, corresponding to 6 degrees of freedom. The control dimension is distinguished by 0 and 1, with 0 controlling position and 1 controlling force. Because this is a grinding operation, position control should be performed parallel to the contact surface and the end effector should not rotate horizontally. In other words, the main diagonal elements of the first, second, fourth, fifth, and sixth rows of the matrix are selected to be 0, and the main diagonal element of the third row is 1.

[0019] In combination with the above method, an embodiment of the present invention provides an implementation method of the self-state prediction model proposed in the above method, which includes: establishing a dynamic model for the robot based on the Lagrangian method; the dynamic model includes the robot's acceleration product term, velocity product term, position product term, feedforward torque term and the robot's gravity compensation term; based on the expected input, according to the constructed selection matrix, the control increment of the corresponding joint space is obtained; according to the self-state prediction model, a force / position augmented state matrix is ​​generated.

[0020] In combination with the above method, an embodiment of the present invention provides an implementation method for superimposing the expected control quantity weights in the above method, which includes: based on the expected input, according to the constructed selection matrix, obtaining the control increment of the corresponding joint space; representing the robot's own system with a second-order impedance model equation; setting the weight of the corresponding control quantity according to the robot's own impedance model; multiplying the control quantity and the corresponding weight and superimposing them into the inertia matrix term of the robot's dynamic equation to achieve simultaneous control of force and position in the corresponding joint space.

[0021] In combination with the above method, an embodiment of the present invention provides a force control implementation method in the above method. The force control strategy adopts a feedforward compensation and gain superposition method. The method includes: in the force control subspace decoupled by the selection matrix, the actual force is compared with the expected force to obtain a control increment, and the product of the weight and the control increment is incorporated into the inertia matrix product term to realize force control. It should be noted that the actual force can be calculated through the robot's own state prediction model.

[0022] In combination with the above method, an embodiment of the present invention provides a position control implementation method in the above method. The position control strategy adopts PD control and gain superposition, which includes: in the position space decoupled by the selection matrix, the actual position is compared with the expected position to obtain a position increment, the position increment is differentiated to obtain a speed control increment, the corresponding coefficient is multiplied by the speed term and the position term, and the weighted form is superimposed and incorporated into the inertia matrix product term to realize position control.

[0023] It should be noted that the actual position can be calculated using the robot's own state prediction model.

[0024] Compared with the current common force / position hybrid control method, this embodiment has the following outstanding advantages.

[0025] In this embodiment, the force control subspace and the position control subspace are weighted to achieve simultaneous force and position control. This avoids the strong decoupling of the selection matrix subspace division and significantly reduces force error fluctuations and position errors. Force-position hybrid control is often prone to dynamic hysteresis when polishing complex surfaces due to the strong decoupling of the selection matrix. This manifests as large force and position errors, and cannot guarantee the polishing effect.

[0026] In this embodiment, the weight parameters are adaptively adjusted, providing greater universality for complex working conditions. The absolute values ​​of force and position errors serve as the inputs to the fuzzy controller, and the three parameter adjustments correspond to the controller's outputs. These adjustments are used to update the core parameters at the next moment, enabling real-time adaptive adjustment and improving the robustness and compliance of the system. Currently, however, the weight settings must be determined based on the system impedance model before the algorithm is established, and cannot be adaptively changed during operation.

[0027] A robot is an industrial robot grinding and polishing compliance fuzzy control system based on weighted force / position superposition control, characterized by comprising:

[0028] The expected input module is configured to input the expected force and expected position of the robot as the expected input of the entire system process;

[0029] The core controller module is configured to calculate the joint output torque by superimposing the force control subinterval and the position control subspace weights using the controller core formula as the control rate;

[0030] The parameter self-tuning module is configured to adopt fuzzy control means, with the absolute value of force control error and the absolute value of position error as the input of the fuzzy controller, and the adjustment amount of the three core parameters in the control law as the output, to achieve parameter self-tuning optimization and improve the adaptability and flexibility of the controller.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] (1) The control method designed in the present invention provides a method for weighted superposition of the force control subspace and the position control subspace, which significantly reduces the dynamic hysteresis problem under high dynamic conditions and can achieve better force tracking and position control effects;

[0033] (2) The control method designed in the present invention can dynamically adjust the gain coefficient of the weighted hybrid control according to the real-time information of the force error and position error through the fuzzy reasoning mechanism, thereby realizing the adaptive optimization of the gain coefficient under different working conditions and improving the compliance of the grinding system;

[0034] (3) The control method and system designed by the present invention provide a good idea of ​​force and position control effect, which can achieve smooth grinding of complex curved surface workpieces. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 Flowchart of the fuzzy control method for the compliance force of industrial robot grinding and polishing based on weighted force / position superposition control provided by an example of the present invention;

[0036] Figure 2The membership function of the input variable of the fuzzy controller provided by the embodiment of the present invention;

[0037] Figure 3 The membership function of the output variable of the fuzzy controller provided by the embodiment of the present invention;

[0038] Figure 4 A control block diagram of a fuzzy control method for the compliance force of an industrial robot grinding and polishing based on weighted force / position superposition control provided by an example of the present invention;

[0039] Figure 5 The fuzzy control block diagram provided for the example of the present invention;

[0040] Figure 6 Position error simulation result diagram provided for the example of the present invention;

[0041] Figure 7 This is a simulation result diagram of the expected force tracking effect under high dynamic conditions provided by an example of the present invention;

[0042] Figure 8 This is a simulation result diagram of the expected force error under high dynamic conditions provided by an example of the present invention;

[0043] Figure 9 This is a simulation result diagram of the expected force tracking effect under constant force grinding conditions provided by an example of the present invention;

[0044] Figure 10 This is a simulation result diagram of the expected force error under constant force grinding conditions provided by an example of the present invention. DETAILED DESCRIPTION

[0045] The specific implementation methods of the present invention are further explained in detail below through description of embodiments with reference to the accompanying drawings, so as to help those skilled in the art to have a more complete, accurate and in-depth understanding of the inventive concept and technical solution of the present invention.

[0046] Figure 1 The flow chart of the fuzzy control method for the compliance force of industrial robot grinding and polishing based on weighted force / position superposition control provided by the embodiment of the present invention includes the following steps:

[0047] Step 1: Establish a machine dynamics model;

[0048]

[0049] Where τ is the joint torque; M(q) is the inertia matrix; q is the joint angle; is the joint angular velocity; is the joint angular acceleration; is the Coriolis force and centrifugal force matrix; G(q) is the gravity matrix.

[0050] The dynamic model of the contact between the robot and the workpiece in the task space is further generalized.

[0051]

[0052] Step 2: Establish a force control selection matrix to decouple the robot's joint space into a force control subspace and a position control subspace;

[0053] The force control subspace and the position control subspace are decoupled and divided by the selection matrix; the selection matrix is ​​expressed as follows:

[0054]

[0055] The matrix is ​​selected to be a 6-dimensional matrix corresponding to 6 degrees of freedom, with the control dimension distinguished by 0 and 1. A value of 0 controls position, while a value of 1 controls force. Because this is a grinding operation, position control should be performed parallel to the contact surface and the end effector should not rotate horizontally. In other words, the main diagonal elements of the first, second, fourth, fifth, and sixth rows of the matrix are selected to be 0, and the main diagonal element of the third row is 1.

[0056] Step 3: Establish a fuzzy controller to self-tune the weight coefficient.

[0057] As shown in Table 1, a fuzzy tuning rule table is established, with the absolute values ​​of force error and position error as the input of the fuzzy controller. Figure 2 The membership function shown in the figure is used to fuzzify the period. Figure 3 The membership function of the output variable shown in the figure obtains the fuzzy result of the output variable; the centroid method is used to defuzzify the result to obtain the adjustment amount of the weight coefficient, thereby realizing the self-tuning of the coefficient.

[0058] Table 1 Fuzzy self-tuning rules for gain coefficient

[0059] Step 4: Obtain the expected force control value, position control value, and joint velocity control value of each joint according to the selection matrix;

[0060] The corresponding expected control quantities are superimposed with weights to achieve simultaneous control of force and position, and obtain the control increment of the robot;

[0061]

[0062] Where I is the six-dimensional identity matrix; and They are velocity gain coefficient, position gain coefficient and force gain coefficient respectively; is the Coriolis force and centrifugal force matrix; G(X) is the gravity matrix; and is the adjustment amount at time i+1; τ i+1 is the joint torque output at time i+1.

[0063] Step 5: According to the control increment, determine the joint torque and joint angle of the robot, and control and drive each joint of the robot so that the robot can achieve the desired input. From formula (3), we can see that the force, position and speed correspond to and The weighted superposition is incorporated into the product term of the inertia matrix to achieve simultaneous control of force and position. At the same time, it also reduces the subspace decoupling of the selection matrix, which can reduce dynamic hysteresis when dealing with complex surface grinding.

[0064] Based on the above, combined Figure 4 The control block diagram of the method proposed in this embodiment further explains the method proposed in this embodiment

[0065] This method proposes to include the force control feedback term into the inertia matrix product term and to perform weighted mixing. That is, in the control block diagram, force control and position control no longer distinguish between inner loop and outer loop, and the feedback is weighted mixed as a supplementary quantity for the torque and position update at the next moment.

[0066] At the same time, this embodiment adopts the means of fuzzy controller to adjust the three gain coefficients K p , K d , K f Self-tuning adjustment. Among them, the process of the fuzzy controller is as follows Figure 5 shown.

[0067] Combined with the design of the weight / position hybrid controller, the contact force error |ΔF| and position error |ΔX| of the robot end effector are used as the input of the fuzzy controller, and Δk p , Δk d , Δk f The three output results are mainly obtained by reasoning based on the set fuzzy rules.

[0068] The above method was simulated in the MATLAB programming environment and the results were as follows: Figures 6-10 The simulation results are shown.

[0069] In position tracking, the position error is obtained by subtracting the actual position from the expected position, which corresponds to the following: Figure 6 Position error graph shown. Figure 6 is the error between the actual position and the expected position of the end effector, Figure 6 The position error control effect of the control method of the present invention can be obtained.

[0070] From the first peak of 1.734mm to the fifth peak of 1.351mm, it gradually decreases and converges, and the average position error is 0.890mm. In terms of position tracking effect, the maximum position error of the fuzzy weighted / bit hybrid control method does not exceed 2mm.

[0071] , the average position error does not exceed 1mm, and the position control effect is good.

[0072] In order to demonstrate that the dynamic hysteresis of the method proposed in this embodiment is reduced under high dynamic conditions, the expected force is set to be simulated by a sinusoidal signal, and the actual force tracking effect is examined, and the following results are obtained: Figure 7 and Figure 8 The simulation results are shown.

[0073] like Figure 7 As shown in the figure, the dashed line represents the actual force signal, and the solid line represents the desired force signal. The proposed method exhibits minimal dynamic hysteresis, primarily manifested by a reduction in the number of actual force "spikes" corresponding to the desired force peak, resulting in a flatter, less dynamic hysteresis. Furthermore, force tracking in the transition section accurately tracks the desired force, ensuring optimal machining results.

[0074] like Figure 8 As shown in the figure, the maximum positive fluctuation value is 0.141 N, the maximum negative fluctuation value is 0.140 N, and the overall maximum fluctuation is 0.280 N. According to the results, the force tracking effect is better under high dynamic conditions.

[0075] Constant force grinding is the mainstream working condition at present. In order to verify the usability of the method proposed in this embodiment, the expected force is set to simulate the step signal and the actual force tracking effect is examined. The following results are obtained: Figure 9 and Figure 10 The simulation results are shown.

[0076] like Figure 9 As shown, the solid line is the actual force signal, and the dotted line is the expected force signal. The force error of the method proposed in this embodiment is small, which is mainly reflected in the expected force tracking. The corresponding fluctuation range is small, the force tracking effect is good, and the processing effect is guaranteed.

[0077] like Figure 10 As shown in the figure, the maximum positive fluctuation value is 0.138 N, the maximum negative fluctuation value is 0.146 N, and the overall maximum fluctuation is 0.284 N. According to the results, the force control effect is better under the constant force grinding condition.

[0078] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

[0079] The contents not described in detail in the specification of the present invention belong to the common knowledge of professionals in this field.

Claims

1. A fuzzy control method for the compliance of industrial robot grinding and polishing based on weighted force / position superposition control, characterized in that: The method specifically comprises the following steps: Step 1: Establish a dynamic model of robot contact with workpiece in Cartesian space; Step 2: Establish a selection matrix to decouple the Cartesian space and superimpose the weights: (1) Calculate the curvature of the wheel hub end face and establish the robot selection matrix; (2) Decoupling the Cartesian space into position control subspace and force control subspace according to the selection matrix; (3) The two control subspaces are superimposed and multiplied into the inertia matrix product term in the form of weights; Step 3: Use fuzzy control to self-tune the coefficients: (1) The absolute values ​​of force error and position error are used as inputs of the fuzzy controller; (2) Obtain the adjustment amount of the core parameter at the next moment; (3) Parameters are optimized through self-tuning; Step 4: Obtain the expected control amount of force and position of each joint; Step 5: Drive and control the robot.

2. The fuzzy control method for industrial robot grinding and polishing compliance based on weighted force / position superposition control according to claim 1 is characterized in that: First, the Lagrangian method is used to establish the dynamic model of the robot in Cartesian space. The model formula is as follows: Where, τ is the joint torque; M(q) is the inertia matrix; q, and are joint angle, joint angular velocity and joint angular acceleration respectively; is the Coriolis force and centrifugal force matrix; G(q) is the gravity matrix; Thirdly, when the end effector is subject to external environmental constraints, that is, when the end effector contacts the workpiece and a force is applied, the dynamic model of the robot is shown in formula (2); Where, τ t is the joint output torque; At the same time, there are t t =J T F t (3) Where J is the Jacobian matrix; F t is the actual force; Finally, the dynamic model of the contact between the robot and the workpiece in the task space is obtained by substituting formula (3) into formula (2). The dynamic model of the robot in the task space can be obtained.

3. The fuzzy control method for industrial robot grinding and polishing compliance based on weighted force / position superposition control according to claim 1 is characterized in that: A selection matrix is ​​established to decouple the Cartesian space. The specific form of the selection matrix is ​​shown in formula (5): It should be noted that the selection matrix is ​​a 6-dimensional matrix corresponding to 6 degrees of freedom, and the control dimension is distinguished by 0 and 1. When it is 0, it controls the position, and vice versa, when it is 1, it controls the force. Due to the grinding operation, the position control should be carried out in the direction parallel to the contact surface and the end effector should not rotate horizontally, that is, the main diagonal elements of the first, second, fourth, fifth and sixth rows in the selection matrix are 0, and the main diagonal element of the third row is 1.

4. The fuzzy control method for industrial robot grinding and polishing compliance based on weighted force / position superposition control according to claim 1 is characterized in that: Formula (6) is used to perform weight superposition and incorporate it into the inertia matrix product term to achieve simultaneous control of force and position; 5. The fuzzy control method for industrial robot grinding and polishing compliance based on weighted force / position superposition control according to claim 1 is characterized in that: The fuzzy controller is constructed; a fuzzy tuning rule table is established, with the absolute values ​​of the force error and position error as the inputs to the fuzzy controller. Because the output varies with time, a triangular membership function is selected. The period is fuzzified using the triangular membership function, and the fuzzified output variable is obtained using the triangular membership function of the output variable. The centroid method is used to defuzzify the weight coefficient and obtain the adjustment value, thus achieving self-tuning of the coefficient.

6. Using the result of formula (6) as the control quantity and the robot's own state prediction model, the control increment of the corresponding joint is obtained to achieve the expected driving control effect of the robot.

7. A robot is an industrial robot grinding and polishing compliance fuzzy control system based on weighted force / position superposition control, characterized in that: include: The expected input module is configured to input the expected force and expected position of the robot as the expected input of the entire system process; The core controller module is configured to calculate the joint output torque by superimposing the force control subinterval and the position control subspace weights using the controller core formula as the control rate; The parameter self-tuning module is configured to adopt fuzzy control means, with the absolute value of force control error and the absolute value of position error as the input of the fuzzy controller, and the adjustment amount of the three core parameters in the control law as the output, to achieve parameter self-tuning optimization and improve the adaptability and flexibility of the controller.

Images