Active milling vibration suppression method and system considering robot posture-related modes

Abstract

The present invention belongs to but is not limited to the field of robot technology, and in particular relates to a method and system for actively suppressing milling vibrations that considers robot posture-related modes. The mode of the robot changes with the change of the robot's posture, so the traditional vibration suppression method is difficult to achieve stable vibration suppression in the robot's large workspace. In order to be able to stably suppress the vibration during the robot's milling process, it is necessary to consider the robot's posture-related modes, adaptively change the relevant parameters of the vibration suppression algorithm, and achieve a stable vibration suppression effect in the robot's large workspace. The present invention constructs a robot modal observer based on a Gaussian regression process to achieve modal parameter observation and state equation update at different robot postures, and designs a linear quadratic regulator with real-time update of the optimal gain of a closed-loop system based on the modal parameters of the robot at different postures, and suppresses the vibration during the robot's milling process through a vibration suppressor.

Description

Technical Field

[0001] The present invention belongs to but is not limited to the field of robot technology, and in particular relates to a method and system for actively suppressing milling vibration by considering robot posture-related modes. Background Art

[0002] At present, in the field of robot milling, the serial structure of the robot leads to low stiffness of the robot end, which is easy to produce large vibration under the action of time-varying cutting force, seriously affecting the machining accuracy and surface quality of robot milling, and restricting the development of robot milling. The existing vibration suppression methods are mainly divided into passive, semi-active and active vibration suppression methods. Passive and semi-active vibration suppression methods are mainly achieved by changing the structure of the mechanism or adding corresponding dampers to the mechanism. Active vibration suppression methods achieve vibration suppression during milling by controlling the behavior of the vibration suppressor. However, for a six-degree-of-freedom serial robot, the modal distribution of the robot is heavily dependent on the robot's posture. Therefore, traditional vibration suppression methods are difficult to achieve stable vibration suppression in the robot's large workspace. In order to stably suppress the vibration during the robot milling process, it is necessary to consider the robot's posture-related modes, adaptively change the relevant parameters of the vibration suppression algorithm, and achieve a stable vibration suppression effect in the robot's large workspace.

[0003] In view of the above analysis, the technical problems that need to be solved urgently in the existing technology are: the existing active vibration suppression methods for robot milling processing have certain limitations and do not consider the robot's mode changes with posture. Under different robot postures and modes, the robot vibration performance generated by the same cutting force varies greatly, and a stable vibration suppression effect cannot be achieved in the robot's large workspace, which limits the promotion and application of robot milling processing.

[0004] The difficulty of solving the above problems and defects is:

[0005] When the robot's posture changes, the robot's modal response also changes. It is very difficult to accurately calculate the robot's modal parameters online. When the robot's modal changes, the traditional vibration suppression algorithm fails, and in severe cases, it amplifies the vibration during the milling process.

[0006] The significance of solving the above problems and defects is:

[0007] This paper discloses an active vibration suppression method considering robot posture-related modes in robot milling. A robot modal observer based on Gaussian regression process is constructed to realize modal parameter observation and state equation update under different robot postures. A linear quadratic regulator with real-time update of the optimal gain of the closed-loop system is designed according to the modal parameters under different robot postures. The vibration in the robot milling process is suppressed by a vibration suppressor. Summary of the invention

[0008] In view of the problems existing in the prior art, the present invention provides a method and system for actively suppressing milling vibration taking into account the robot's posture-related modes, constructs a robot modal observer based on a Gaussian regression process, realizes modal parameter observation and state equation updating in different robot postures, and designs a linear quadratic regulator that updates the optimal gain of a closed-loop system in real time according to the modal parameters in different robot postures. The vibration in the robot's milling process is suppressed by a vibration suppressor, thereby improving the machining accuracy and stability of the robot during milling.

[0009] The present invention is implemented as follows: a method for actively suppressing milling vibration considering robot posture-related modes, comprising:

[0010] S1, evenly divide the robot workspace, attach the accelerometer to the tip of the robot, and perform a modal impact test with a hammer to obtain the modal parameters of the robot at different positions in the workspace;

[0011] S2, establish a Gaussian regression prediction model to predict the first-order natural frequency w of the robot in different positions n , the first-order modal mass M, the first-order damping ratio ξ, the joint angles measured using the modal knocking test in S1 and the first-order natural frequency w of the robot n , the first-order modal mass M, the first-order damping ratio ξ are used to train the Gaussian regression prediction model, and the trained model is used to estimate the modal parameters of the robot in other positions;

[0012] S3, establish an online robot modal observer based on the trained Gaussian process regression model, predict the modal parameters of the robot at the current position by reading the robot's joint angles online, and update the state equation of the robot at the current position;

[0013] S4, design a linear quadratic regulator to control the system, and design the corresponding optimal linear feedback controller gain update strategy, calculate the optimal control gain and input under different robot postures, and suppress the vibration of the robot during milling through a vibration suppressor to achieve a stable milling vibration suppression effect of the robot in a large workspace;

[0014] S5, installing the vibration suppressor to the corresponding robot end, and controlling the input of the vibration suppressor through the milling vibration active suppression method, thereby reducing the vibration during the robot milling process and improving the robot milling accuracy and surface quality.

[0015] Furthermore, S1 specifically includes: modal parameters of the robot at different positions in the workspace, including the first-order natural frequency w n, first-order modal mass M, first-order damping ratio ξ, and based on this, the dynamic model and state space equation of the robot at different positions are established;

[0016] The dynamic model between the acceleration and input force of the robot at a certain position can be expressed as:

[0017]

[0018] Further, S2 specifically includes: assuming that the measured sample set {f(x)} satisfies

[0019] f(x)~GP(μ(x),C(x,x * ))(2)

[0020] Where μ(x) is the basis function, x includes the joint angles corresponding to all robot poses in the training set, and C(x,x * ) represents the covariance matrix of the training sample, and the square exponential kernel function is used to calculate the covariance matrix f(x i ,x j )

[0021]

[0022] where σ l represents the characteristic length, σ f represents the standard deviation of the signal, x i and x j represents the robot joint angle position vector under the corresponding posture, then the corresponding prior prediction result of the predicted position is:

[0023]

[0024] where σ 2 represents the model equation, I represents the identity matrix, x represents the robot joint positions in all training sets, and x * represents the joint positions in the test set, C(x,x), C(x * ,x) and C(x * ,x * ) are the covariance matrix of the training sample, the covariance matrix of the training sample and the test sample, and the covariance matrix of the test sample. The posterior distribution and mean predicted by Bayesian inference can be expressed as:

[0025] f(x * )=μ(x * )+C(x * ,x)[C tr (x,x)+σ 2 I] -1 (f tr (x)-μ tr) (5)

[0026] where f tr (x) and μ tr Represent the value of the prediction parameter and its mean, respectively.

[0027] Further, S3 specifically includes: The state space equation of the robot's dynamic model can be expressed as:

[0028]

[0029] Where u is the input force of the external vibration suppressor on the robot, and the output y is the robot acceleration corresponding to the input force. The state matrix is:

[0030]

[0031] Further, S4 specifically includes: according to the state space equation of the robot, establishing a corresponding cost function as follows:

[0032]

[0033] Q and R are diagonal matrices, representing the penalty intensity of the robot's current state x and control variable u respectively. The larger the value of Q or R, the greater the penalty intensity and the higher the importance of the variable. It is a continuous differential equation, and the robot in the control process is a discrete point, which needs to be discretized. According to the mean value theorem of integral, it can be simplified to:

[0034]

[0035] According to the discretized constraint equation, the objective function can be obtained as follows:

[0036]

[0037] According to the Lagrange multiplier method, the corresponding cost function can be constructed:

[0038]

[0039] The final control quantity is:

[0040]

[0041] In the above formula, only λ k+1 is the unknown quantity, combining the above formulas to get λ k =2P k x k , where P k According to the Riccati equation:

[0042]

[0043] Therefore, the final LQR control feedback control quantity u k It can be expressed as:

[0044]

[0045] Since the robot's mode will change with the posture, the corresponding state space equation matrices A and B will also change. Therefore, when the robot's state space equation changes, the feedback gain K needs to be changed accordingly to achieve the optimal vibration suppression effect under different robot postures.

[0046] Another object of the present invention is to provide a system for actively suppressing milling vibrations by considering robot posture-related modes, which implements the method for actively suppressing milling vibrations by considering robot posture-related modes, comprising:

[0047] The modal parameter acquisition module evenly divides the robot's workspace, attaches an accelerometer to the tip of the robot's tool, and performs a modal impact test with a hammer to obtain the robot's modal parameters at different positions in the workspace.

[0048] Modal parameter estimation module, establishes a Gaussian regression prediction model to predict the first-order natural frequency w of the robot in different positions n , the first-order modal mass M, the first-order damping ratio ξ, the joint angles measured using the modal knocking test in S1 and the first-order natural frequency w of the robot n , the first-order modal mass M, the first-order damping ratio ξ are used to train the Gaussian regression prediction model, and the trained model is used to estimate the modal parameters of the robot in other positions;

[0049] The state equation update module establishes an online robot modal observer based on the trained Gaussian process regression model, predicts the modal parameters of the robot at the current position by reading the robot's joint angles online, and updates the state equation of the robot at the current position.

[0050] The vibration suppression module designs a linear quadratic regulator to control the system, and designs the corresponding optimal linear feedback controller gain update strategy, calculates the optimal control gain and input under different robot postures, and suppresses the vibration of the robot during milling through a vibration suppressor, achieving a stable milling vibration suppression effect of the robot in a large workspace;

[0051] The vibration suppressor installation module installs the vibration suppressor to the corresponding robot end, and controls the input of the vibration suppressor through the active milling vibration suppression method, thereby reducing the vibration during the robot milling process and improving the robot milling accuracy and surface quality.

[0052] Another object of the present invention is to provide a computer device, the computer device includes a memory and a processor, the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method for active suppression of milling vibration considering the robot posture-related modes.

[0053] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to execute the steps of the method for actively suppressing milling vibration taking into account robot posture-related modes.

[0054] Another object of the present invention is to provide an information data processing terminal, which includes the milling vibration active suppression system that takes into account the robot posture-related modes.

[0055] In combination with the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solutions to be protected by the present invention are as follows:

[0056] First, the present invention designs a robot modal observer based on Gaussian regression process, which can train Gaussian regression process model according to a small amount of force hammer knocking modal data, and quickly calculate the robot modal parameters according to the trained Gaussian regression process model and the joint angle of the robot in the current posture, so as to realize the modal parameter observation and state equation update of the robot in different postures.

[0057] The present invention designs a linear quadratic regulator that updates the optimal gain of a closed-loop system in real time according to the modal parameters of the robot in different positions and postures, calculates the optimal control gain and input in different positions and postures of the robot, and suppresses the vibration of the robot during milling by a vibration suppressor, thereby achieving a stable milling vibration suppression effect of the robot in a large working space.

[0058] The present invention can effectively reduce the milling vibration of the robot in a large working space, solve the influence caused by the robot's posture-related modes, avoid the influence of vibration during robot milling on the robot's milling accuracy and stability, and can effectively promote the application and development of robot milling.

[0059] Second, the expected benefits and commercial value of the technical solution of the present invention after transformation are:

[0060] By actively suppressing vibrations during the milling process, the present invention can significantly improve the accuracy and surface quality of robot milling processing. This improvement has great commercial value for industries with high requirements for high-precision processing, such as aerospace manufacturing. High-quality processed parts can reduce the cost of subsequent processing and adjustment, and improve product consistency and reliability. At the same time, the technical solution of the present invention is applicable to a variety of robot processing scenarios, not only limited to milling processing, but also can be applied to other high-precision processing fields, such as turning, drilling, grinding, etc. Its wide applicability makes the present invention have great market potential and can meet the diverse needs of different customers.

[0061] The technical solution of the present invention overcomes technical prejudice: traditional methods usually assume that the modal parameters of the system (such as natural frequency, modal mass and damping ratio) are fixed under different working conditions. However, in practical applications, the modal parameters of the robot will change with its posture. Ignoring this will lead to poor vibration suppression effect. Traditional methods are effective in ideal environments, but in complex actual processing environments, the vibration suppression effect is often greatly reduced due to various factors. The present invention uses a Gaussian regression prediction model and a modal observer to observe and update the modal parameters of the robot in different postures in real time, fully considering the modal changes related to the robot's posture, and adapting to complex processing environments by updating the control strategy in real time, thus overcoming the shortcomings of traditional methods in complex environments.

[0062] Third, the present invention provides a method for actively suppressing milling vibrations that takes into account robot posture-related modes, aiming to improve the accuracy and surface quality of robot milling processing. First, in step S1, the robot workspace is evenly divided, and an accelerometer is attached to the tip of the robot tool to perform a force hammer modal impact test to obtain the modal parameters of the robot in different postures, such as the first-order natural frequency, first-order modal mass, and first-order damping ratio. These parameters provide basic data for the subsequent establishment of dynamic models and state-space equations.

[0063] Next, in step S2, the modal parameters of the robot in different positions are predicted using the Gaussian regression prediction model. The joint angles and modal parameters measured by the modal knocking experiment are used as training data to train the Gaussian regression prediction model, and then the modal parameters of the robot in other positions are estimated. This prediction model can effectively cope with the changes in the modal parameters of the robot in different positions, laying the foundation for real-time modal monitoring.

[0064] In step S3, an online robot modal observer is established based on the trained Gaussian process regression model. By reading the robot's joint angles in real time, the observer can predict the robot's modal parameters at the current position and update the robot's state equation at the current position. This online modal monitoring mechanism ensures that the robot can obtain accurate modal parameters under different operating conditions, ensuring the effectiveness of vibration suppression control.

[0065] Finally, in steps S4 and S5, a linear quadratic regulator is designed to control the system, and an optimal linear feedback controller gain update strategy is designed to calculate the optimal control gain and input under different postures. The vibration suppressor is installed at the end of the robot to suppress the vibration during milling by adjusting the controller input in real time. This method achieves a stable milling vibration suppression effect in a large workspace by performing optimal control under different postures of the robot, significantly improving the accuracy and surface quality of milling processing. This active suppression method solves the problem of modal parameter changes caused by posture changes in the prior art, showing significant technological progress.

[0066] Fourth, traditional robot milling vibration suppression methods often do not consider the changes in the modal parameters of the robot in different positions, resulting in limited vibration suppression effects and difficulty in achieving stable vibration suppression in a large workspace. Especially when the robot's position changes, due to the change in modal parameters, traditional fixed parameter vibration suppressors often cannot adapt to such changes, resulting in a decrease in suppression effect.

[0067] The present invention establishes the dynamic model and state space equation of the robot at different positions by introducing parameters that consider the robot's position-related modes, including the first-order natural frequency, the first-order modal mass and the first-order damping ratio. These parameters can accurately reflect the vibration characteristics of the robot at different positions and provide an accurate mathematical model for subsequent vibration suppression.

[0068] The present invention adopts Gaussian regression prediction model to predict the modal parameters of the robot in different positions and postures, and uses these parameters to establish an online robot modal observer to update the state equation of the robot in the current position and posture in real time. Combined with the linear quadratic regulator (LQR), the optimal linear feedback controller is designed to achieve optimal vibration suppression in different positions and postures. This vibration suppression algorithm based on real-time modal parameters significantly improves the accuracy and adaptability of vibration suppression.

[0069] The present invention has made significant technical progress in solving the technical problems existing in the traditional robot milling vibration suppression. The parametric modeling method considering the robot posture-related modes, the modal parameter prediction algorithm based on Gaussian regression prediction, and the optimal vibration suppression strategy combined with LQR together constitute a set of efficient and accurate robot milling vibration active suppression solutions. This solution not only improves the accuracy and surface quality of robot milling processing, but also provides strong support for the stable milling of the robot in a large workspace. The present invention has broad application prospects and social and economic benefits in the fields of robot processing, precision manufacturing, etc. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 is a flow chart of an active vibration suppression method considering robot posture-related modes provided by an embodiment of the present invention;

[0071] Figure 2 is a block diagram of active vibration suppression control considering robot posture-related modes provided by an embodiment of the present invention;

[0072] Figure 3 It is a schematic diagram of actual installation and application of the active vibration suppression method considering robot posture-related modes provided by an embodiment of the present invention in robot milling processing;

[0073] Figure 4 It is a structural diagram of an active vibration suppression system considering robot posture-related modes provided by an embodiment of the present invention.

[0074] Figure 5 It is a robot terminal motion trajectory diagram for verifying the effect of posture-changing vibration suppression provided by an embodiment of the present invention;

[0075] Figure 6 is a rendering of a vibration suppression method considering changes in robot posture provided by an embodiment of the present invention; DETAILED DESCRIPTION

[0076] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0077] like Figure 1 , Figure 2 The present invention is implemented as follows: an active vibration suppression method considering robot posture-related modes in robot milling, specifically comprising the following steps:

[0078] S1, the robot workspace is evenly divided, and the accelerometer is attached to the tip of the robot. The modal knocking test is performed by a force hammer to obtain the modal parameters of the robot at different positions in the workspace, including the first-order natural frequency wn , the first-order modal mass M, the first-order damping ratio ξ, and based on this, the dynamic model and state-space equations of the robot at different positions are established.

[0079] The dynamic model between the acceleration and input force of the robot at a certain position can be expressed as:

[0080]

[0081] S2, establish a Gaussian regression prediction model to predict the first-order natural frequency w of the robot in different positions n , the first-order modal mass M, the first-order damping ratio ξ, the joint angles measured using the modal knocking test in S1 and the first-order natural frequency w of the robot n , the first-order modal mass M, and the first-order damping ratio ξ are used to train the Gaussian regression prediction model, and the trained model is used to estimate the modal parameters of the robot in other positions.

[0082] Assume that the measured sample set {f(x)} satisfies

[0083] f(x)~GP(μ(x),C(x,x * ))(2)

[0084] Where μ(x) is the basis function, x includes the joint angles corresponding to all robot poses in the training set, and C(x,x * ) represents the covariance matrix of the training sample, and the square exponential kernel function is used to calculate the covariance matrix f(x i ,x j )

[0085]

[0086] where σ l represents the characteristic length, σ f represents the standard deviation of the signal, x i and x j represents the robot joint angle position vector under the corresponding posture, then the corresponding prior prediction result of the predicted position is:

[0087]

[0088] where σ 2 represents the model equation, I represents the identity matrix, x represents the robot joint positions in all training sets, and x * represents the joint positions in the test set, C(x,x), C(x * ,x) and C(x * ,x *) are the covariance matrix of the training sample, the covariance matrix of the training sample and the test sample, and the covariance matrix of the test sample. The posterior distribution and mean predicted by Bayesian inference can be expressed as:

[0089] f(x * )=μ(x * )+C(x * ,x)[C tr (x,x)+σ 2 I] -1 (f tr (x)-μ tr ) (5)

[0090] where f tr (x) and μ tr Represent the value of the prediction parameter and its mean, respectively.

[0091] S3, based on the trained Gaussian process regression model, establishes an online robot modal observer, predicts the modal parameters of the robot at the current position by reading the robot's joint angles online, and updates the state equation of the robot at the current position. More specifically, let The state space equation of the robot's dynamic model can be expressed as:

[0092]

[0093] Where u is the input force of the external vibration suppressor on the robot, and the output y is the robot acceleration corresponding to the input force. The state matrix is:

[0094]

[0095] S4, design a linear quadratic regulator to control the system, and design the corresponding optimal linear feedback controller gain update strategy, calculate the optimal control gain and input under different robot postures, and suppress the vibration of the robot during milling through a vibration suppressor to achieve a stable milling vibration suppression effect of the robot in a large workspace. More specifically, according to the state space equation of the robot, the corresponding cost function is established as:

[0096]

[0097] Q and R are diagonal matrices, representing the penalty intensity of the robot's current state x and control variable u respectively. The larger the value of Q or R, the greater the penalty intensity and the higher the importance of the variable. It is a continuous differential equation, and the robot in the control process is a discrete point, which needs to be discretized. According to the mean value theorem of integral, it can be simplified to:

[0098]

[0099] According to the discretized constraint equation, the objective function can be obtained as follows:

[0100]

[0101] According to the Lagrange multiplier method, the corresponding cost function can be constructed:

[0102]

[0103] The final control quantity is:

[0104]

[0105] In the above formula, only λ k+1 is the unknown quantity, combining the above formulas to get λ k =2P k x k , where P k According to the Riccati equation:

[0106]

[0107] Therefore, the final LQR control feedback control quantity u k It can be expressed as:

[0108]

[0109] Since the robot's mode will change with the posture, the corresponding state space equation matrices A and B will also change. Therefore, when the robot's state space equation changes, the feedback gain K needs to be changed accordingly to achieve the optimal vibration suppression effect under different robot postures.

[0110] S5, such as Figure 3 As shown, the vibration suppressor is installed at the corresponding end of the robot, and the input of the vibration suppressor is controlled by the active milling vibration suppression method proposed in the present invention, thereby reducing the vibration during the robot milling process and improving the robot milling accuracy and surface quality.

[0111] like Figure 4 As shown, the milling vibration active suppression system considering robot posture-related modes provided by the embodiment of the present invention includes:

[0112] The modal parameter acquisition module evenly divides the robot's workspace, attaches an accelerometer to the tip of the robot's tool, and performs a modal impact test with a hammer to obtain the robot's modal parameters at different positions in the workspace.

[0113] Modal parameter estimation module, establishes a Gaussian regression prediction model to predict the first-order natural frequency w of the robot in different positions n , the first-order modal mass M, the first-order damping ratio ξ, the joint angles measured using the modal knocking test in S1 and the first-order natural frequency w of the robot n , the first-order modal mass M, the first-order damping ratio ξ are used to train the Gaussian regression prediction model, and the trained model is used to estimate the modal parameters of the robot in other positions;

[0114] The state equation update module establishes an online robot modal observer based on the trained Gaussian process regression model, predicts the modal parameters of the robot at the current position by reading the robot's joint angles online, and updates the state equation of the robot at the current position.

[0115] The vibration suppression module designs a linear quadratic regulator to control the system, and designs the corresponding optimal linear feedback controller gain update strategy, calculates the optimal control gain and input under different robot postures, and suppresses the vibration of the robot during milling through a vibration suppressor, achieving a stable milling vibration suppression effect of the robot in a large workspace;

[0116] The vibration suppressor installation module installs the vibration suppressor to the corresponding robot end, and controls the input of the vibration suppressor through the active milling vibration suppression method, thereby reducing the vibration during the robot milling process and improving the robot milling accuracy and surface quality.

[0117] An application embodiment of the present invention provides a computer device, which includes a memory and a processor. The memory stores a computer program. When the computer program is executed by the processor, the processor executes the steps of a method for actively suppressing milling vibrations that takes into account robot posture-related modes.

[0118] An application embodiment of the present invention provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the processor executes the steps of a method for actively suppressing milling vibrations taking into account robot posture-related modes.

[0119] An application embodiment of the present invention provides an information data processing terminal, which includes a milling vibration active suppression system that takes into account robot posture-related modes.

[0120] The present invention is specifically applied to the field of robot processing, and can improve the stability and accuracy of the robot during processing, assembly and handling, reduce the impact of vibration on processing quality, and enhance the applicable field of robot automated processing.

[0121] An eccentric mass block is installed on the electric spindle at the end of the robot. The eccentric mass block is used to simulate the periodic force during milling. In order to simulate the change of the robot's posture during machining, a spiral trajectory covering the robot's workspace is designed to verify the effectiveness of the method of the present invention under different robot postures. The motion trajectory of the robot end in the experimental verification is as follows: Figure 5 .

[0122] The effect of using the vibration suppression method considering the robot posture change proposed by the present invention is as follows: Figure 6 As shown, the vibration suppression method without considering the robot's posture can also play a role in vibration suppression. However, in some robot postures, the vibration suppression effect is not good or even cannot suppress the light at that position. Compared with not considering the robot's posture, the proposed method can maintain a stable vibration suppression effect in the entire workspace and achieve vibration suppression during the processing. Compared with the vibration suppression method without considering the robot's posture, the vibration suppression method proposed by the present invention can effectively reduce the vibration during the processing and improve the accuracy and surface quality of robot processing.

[0123] It should be noted that the embodiments of the present invention can be implemented by hardware, software, or a combination of software and hardware. The hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. It can be understood by a person of ordinary skill in the art that the above-mentioned devices and methods can be implemented using computer executable instructions and / or contained in a processor control code, such as a carrier medium such as a disk, CD or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. Such code is provided on the carrier medium. The device and its modules of the present invention can be implemented by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., can also be implemented by software executed by various types of processors, and can also be implemented by a combination of the above-mentioned hardware circuits and software, such as firmware.

[0124] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principle of the present invention should be covered by the protection scope of the present invention.

Claims

1. A method for actively suppressing milling vibration considering robot posture-related modes, characterized in that: include: S1, evenly divide the robot workspace, attach the accelerometer to the tip of the robot, and perform a modal impact test with a hammer to obtain the modal parameters of the robot at different positions in the workspace; S2, establish a Gaussian regression prediction model to predict the first-order natural frequency w of the robot in different positions n , the first-order modal mass M, the first-order damping ratio ξ, the joint angles measured using the modal knocking test in S1 and the first-order natural frequency w of the robot n , the first-order modal mass M, the first-order damping ratio ξ are used to train the Gaussian regression prediction model, and the trained model is used to estimate the modal parameters of the robot in other positions; S3, establish an online robot modal observer based on the trained Gaussian regression prediction model, predict the modal parameters of the robot at the current position by reading the joint angles of the robot online, and update the state equation of the robot at the current position; S4, design a linear quadratic regulator to control the system, and design the corresponding optimal linear feedback controller gain update strategy, calculate the optimal control gain and input under different robot postures, and suppress the vibration of the robot during milling through a vibration suppressor to achieve a stable milling vibration suppression effect of the robot in a large workspace; S5, installing a vibration suppressor to the corresponding robot end, and controlling the input of the vibration suppressor by an active milling vibration suppression method, thereby reducing the vibration during the robot milling process and improving the robot milling accuracy and surface quality; S3 specifically includes: The state space equation of the robot's dynamic model can be expressed as: Where u is the input force of the external vibration suppressor on the robot, and the output y is the robot acceleration corresponding to the input force. The state matrix is: S4 specifically includes: according to the state space equation of the robot, establishing the corresponding cost function as follows: Among them, Q and R are diagonal matrices, representing the punishment intensity of the robot's current state x and control quantity u respectively. The larger the value of Q or R, the greater the punishment intensity and the higher the degree of attention to the robot's current state x or control quantity u. It is a continuous differential equation, but the robot in the control process is a discrete point, which needs to be discretized. According to the mean value theorem of integral, it can be simplified to: According to the discretized constraint equation, the objective function can be obtained as follows: According to the Lagrange multiplier method, the corresponding cost function can be constructed: The final control quantity is: In the above formula, only λ k+1 is the unknown quantity, combining the above formulas to get λ k =2P k x k , where P k According to the Riccati equation: Therefore, the final LQR control feedback control quantity u k It can be expressed as: Since the robot's mode will change with the posture, the corresponding state space equation matrices A and B will also change. Therefore, when the robot's state space equation changes, the feedback gain K needs to be changed accordingly to achieve the optimal vibration suppression effect under different robot postures.

2. The method for actively suppressing milling vibration considering robot posture-related modes as claimed in claim 1, characterized in that: S1 specifically includes: modal parameters of the robot at different positions in the workspace, including the first-order natural frequency w n , first-order modal mass M, first-order damping ratio ξ, and based on this, the dynamic model and state space equation of the robot at different positions are established; The dynamic model between the acceleration and input force of the robot at a certain position can be expressed as:

3. The method for actively suppressing milling vibration considering robot posture-related modes as claimed in claim 1, characterized in that: S2 specifically includes: Assume that the measured sample set {f(x)} satisfies f(x)~GP(μ(x),C(x,x))(2)where μ(x) is the basis function, x includes the joint angles corresponding to all robot postures in the training set, C(x,x) represents the covariance matrix of the training samples, and the square exponential kernel function is used to calculate the covariance matrix f(x) of the corresponding posture joint angles. i ,x j ) where σ l represents the characteristic length, σ f represents the standard deviation of the signal, x i and x j represents the robot joint angle position vector under the corresponding posture, then the corresponding prior prediction result of the predicted position is: where σ 2 represents the model equation, I represents the identity matrix, x represents the robot joint positions in all training sets, and x * represents the joint positions in the test set, C(x,x), C(x * ,x) and C(x * ,x * ) are the covariance matrix of the training sample, the covariance matrix of the training sample and the test sample, and the covariance matrix of the test sample. The posterior distribution and mean predicted by Bayesian inference can be expressed as: f(x * )=μ(x * )+C(x * ,x)[C tr (x,x)+σ 2 I] -1 (f tr (x)-μ tr ) (5) where f tr (x) and μ tr Represent the value of the prediction parameter and its mean, respectively.

4. A system for actively suppressing milling vibrations by considering robot posture-related modes, which implements the method for actively suppressing milling vibrations by considering robot posture-related modes as claimed in any one of claims 1 to 3, comprising: The modal parameter acquisition module evenly divides the robot's workspace, attaches an accelerometer to the tip of the robot's tool, and performs a modal impact test with a hammer to obtain the robot's modal parameters at different positions in the workspace. Modal parameter estimation module, establishes a Gaussian regression prediction model to predict the first-order natural frequency w of the robot in different positions n , the first-order modal mass M, the first-order damping ratio ξ, the joint angles measured using the modal knocking test in S1 and the first-order natural frequency w of the robot n , the first-order modal mass M, the first-order damping ratio ξ are used to train the Gaussian regression prediction model, and the trained model is used to estimate the modal parameters of the robot in other positions; The state equation update module establishes an online robot modal observer based on the trained Gaussian process regression model, predicts the modal parameters of the robot at the current position by reading the robot's joint angles online, and updates the state equation of the robot at the current position. The vibration suppression module designs a linear quadratic regulator to control the system, and designs the corresponding optimal linear feedback controller gain update strategy, calculates the optimal control gain and input under different robot postures, and suppresses the vibration of the robot during milling through a vibration suppressor, achieving a stable milling vibration suppression effect of the robot in a large workspace; The vibration suppressor installation module installs the vibration suppressor to the corresponding robot end, and controls the input of the vibration suppressor through the active milling vibration suppression method, thereby reducing the vibration during the robot milling process and improving the robot milling accuracy and surface quality.

5. A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method for active suppression of milling vibration considering robot posture-related modes as described in any one of claims 1 to 3.

6. A computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to execute the steps of the method for active milling vibration suppression considering robot posture-related modes as described in any one of claims 1 to 3.

7. An information data processing terminal, comprising the milling vibration active suppression system considering robot posture-related modes as claimed in claim 4.

Images