A heavy-load robot parameter online identification method based on digital twinning

Abstract

The application relates to the technical field of digital twinning, and particularly discloses a heavy-load mechanical arm parameter online identification method based on digital twinning, which comprises the following steps: linearizing a constructed mechanical arm joint inertia equation to separate a mechanical arm joint inertia parameter observation matrix and a to-be-identified mechanical arm joint inertia parameter; linearizing an obtained mechanical arm joint friction force equation to separate a mechanical arm joint friction force parameter observation matrix and a to-be-identified mechanical arm joint friction force parameter; combining the two observation matrices into a mechanical arm joint parameter observation matrix and combining the two parameters into the to-be-identified mechanical arm joint parameter; obtaining a total observation equation of the mechanical arm according to the mechanical arm joint parameter observation matrix and the to-be-identified mechanical arm joint parameter; and substituting screened mechanical arm joint data into the total observation equation of the mechanical arm to calculate the to-be-identified mechanical arm joint parameter. The application can analyze and correct the mechanical arm dynamics parameters online through the mechanical arm upper sensor, so as to facilitate the parameter adjustment of the mechanical arm control algorithm.

Description

Technical Field

[0001] This invention relates to the field of digital twin technology, and more specifically, to a method for online identification of parameters of a heavy-duty robotic arm based on digital twins. Background Technology

[0002] Digital twin technology holds great promise for the research and application of robotic arms, constructing a new system for describing, diagnosing, predicting, and making decisions about them. Parameter identification technology is a crucial component of the virtual-real interaction technology for robotic arms, such as... Figure 1 As shown, this enables digital twin models to more accurately describe and predict the work of robotic arms. Existing parameter identification methods and systems have been researched to some extent in the current academic and technological fields, but from an application perspective, they have the following problems: 1. Existing digital twin systems cannot accurately acquire the parameters of the robotic arm, nor can they accurately capture parameter changes during operation; 2. Existing parameter identification methods require independent design of experimental schemes and excitation trajectories for the robotic arm to test its parameters, and cannot perform real-time online parameter identification, nor can they accurately capture parameter changes caused by the robotic arm's operation.

[0003] For example, the method for identifying the end-effector load dynamic parameters of a robotic arm based on a six-dimensional force sensor, patent number CN 117021109 A, uses a force sensor at the end of the robotic arm to obtain working data and calculate the dynamic parameters of the robotic arm. However, the method requires independent experiments to test the dynamic parameters of the robotic arm and proposes a specific calculation method for the design of the excitation trajectory of the experiment. Therefore, the method is not adaptable to the motion trajectory of the robotic arm and cannot handle the motion data of the robotic arm under common working conditions. Thus, it cannot perform online parameter identification based on the work of the robotic arm. Summary of the Invention

[0004] To address the technical problem that existing robotic arm parameter identification technologies cannot perform online real-time identification, this invention provides a method for online identification of heavy-duty robotic arm parameters based on digital twins, which can perform online parameter identification based on the operation of the robotic arm.

[0005] As a first aspect of the present invention, a method for online identification of heavy-duty robotic arm parameters based on digital twins is provided, comprising the following steps:

[0006] Step S1: Derive the robotic arm dynamics formula according to the Newton-Euler recursive method to construct the robotic arm joint inertial equation, and linearize the robotic arm joint inertial equation to separate the robotic arm joint inertial parameter observation matrix and the robotic arm joint inertial parameters to be identified.

[0007] Step S2: Obtain the joint friction equation of the robotic arm and linearize the joint friction equation of the robotic arm to separate the observation matrix of joint friction parameters of the robotic arm and the joint friction parameters of the robotic arm to be identified;

[0008] Step S3: Merge the observation matrix of the inertial parameters of the robotic arm joint and the observation matrix of the frictional force of the robotic arm joint into a robotic arm joint parameter observation matrix, and at the same time merge the inertial parameters of the robotic arm joint to be identified and the frictional force parameters of the robotic arm joint to be identified into the joint parameters of the robotic arm to be identified.

[0009] Step S4: Obtain the overall observation equation of the robotic arm based on the observation matrix of the robotic arm joint parameters and the joint parameters of the robotic arm to be identified;

[0010] Step S5: Collect robotic arm joint data in real time and filter the collected robotic arm joint data to obtain filtered robotic arm joint data; wherein, the robotic arm joint data includes robotic arm joint motion trajectory data and robotic arm joint load data, and the robotic arm joint load data includes robotic arm joint inertial load data and robotic arm joint friction load data.

[0011] Step S6: Substitute the filtered robotic arm joint data into the overall observation equation of the robotic arm to calculate the joint parameters of the robotic arm to be identified.

[0012] Furthermore, the step of deriving the robotic arm dynamics formula based on the Newton-Euler recursive method to construct the robotic arm joint inertia equation also includes:

[0013] The recursive process of the robotic arm dynamics formula is as follows:

[0014]

[0015] Where, ω or Represents the angular velocity vector. R represents the angular acceleration vector, R represents the rotation matrix, and v represents the velocity vector. The vector represents the acceleration vector, r is the relative position vector between the two joints, z is the direction vector of the z-axis of the robot arm coordinate system, f is the joint load force vector, μ is the joint load torque vector, F is the resultant force on the joint, and I is the joint moment of inertia. The subscript indicates the joint or axis number being described, and the superscript indicates the robot arm coordinate system number. The robot arm coordinate system is created sequentially from the base to the end effector using the DH method, and the homogeneous transformation matrix of each coordinate system is calculated. From this, the rotation matrix and the Z-axis direction vector are extracted from the homogeneous transformation matrix.

[0016] The joint inertia equation of the robotic arm is constructed based on the recursive process of the robotic arm dynamics formula.

[0017] Furthermore, the Newton-Euler recursive method includes two aspects:

[0018] The first aspect: by using the velocity, angular velocity, acceleration, and angular acceleration of the robotic arm joints themselves, the velocity, angular velocity, acceleration, and angular acceleration of each axis of the robotic arm relative to the world coordinate system, as well as the velocity, angular velocity, acceleration, and angular acceleration of each joint of the robotic arm relative to the world coordinate system, are calculated. This process is a recursion from the base of the robotic arm to the end effector of the robotic arm.

[0019] The second aspect is that the load at the end of the robotic arm is used to progressively increase the load on each joint of the robotic arm based on the movement of each axis and joint. This process is a progressive increase from the end of the robotic arm to the base of the robotic arm.

[0020] Furthermore, the linearization of the robotic arm joint inertial equations to separate the robotic arm joint inertial parameter observation matrix and the robotic arm joint inertial parameters to be identified also includes:

[0021] When the robotic arm is a six-axis robotic arm, the equations of inertia for the robotic arm joints are arranged as follows:

[0022]

[0023] Where, τ links This represents the inertial load data of the robotic arm joints, where Z is the zero matrix. For the unprocessed robotic arm joint inertial parameters, This is the unprocessed observation matrix of the robotic arm joint inertial parameters; wherein, the robotic arm joint motion trajectory data is input into the robotic arm joint inertial equation to calculate...

[0024] The minimum inertia parameter of the robotic arm is obtained based on the DH parameters and the joint type of the robotic arm, and this minimum inertia parameter is used as the inertia parameter χ of the joint of the robotic arm to be identified. z The unprocessed robotic arm joint inertial parameters The inertial parameter χ of the joint of the robotic arm to be identified z The conversion relationship between them is as follows:

[0025]

[0026] Wherein, P is the unprocessed robotic arm joint inertial parameter. The inertial parameter χ of the joint of the robotic arm to be identified z The transformation matrix between them;

[0027] According to the transformation formula Transforming the joint inertia equations of the robotic arm, we obtain:

[0028]

[0029] Among them, W z This is the processed observation matrix of the robotic arm joint inertial parameters.

[0030] Furthermore, the process of obtaining the joint friction equation of the robotic arm and linearizing the joint friction equation to separate the observation matrix of the joint friction parameters of the robotic arm and the joint friction parameters to be identified also includes:

[0031] The friction force equations for each joint of the robotic arm are obtained, and the friction force equations for each joint of the robotic arm are as follows:

[0032]

[0033] Where, τ fic,i Let be the frictional load data for the i-th joint of the robotic arm. Let μ be the velocity or angular velocity of the i-th joint of the robotic arm. i , λ i γ i These are the frictional force parameters to be identified for the i-th joint of the robotic arm;

[0034] The friction force equations for each joint of the robotic arm are linearized and expressed as vectors for each joint:

[0035]

[0036] The vectors of each joint of the robotic arm are arranged to obtain the joint friction force equation of the robotic arm, wherein the joint friction force equation of the robotic arm is as follows:

[0037]

[0038] Where, τ fric For the joint friction load data of the robotic arm, χ f To identify the frictional parameters of the robotic arm joints, W f This is the observation matrix for the joint friction parameters of the robotic arm; wherein, the motion trajectory data of the robotic arm joints is input into the joint friction equation of the robotic arm to calculate W. f .

[0039] Furthermore, the step of obtaining the overall observation equation for the robotic arm based on the observation matrix of the robotic arm joint parameters and the joint parameters of the robotic arm to be identified also includes:

[0040] The joint inertia equation and the joint friction equation of the robotic arm are combined to obtain the overall observation equation of the robotic arm, which is as follows:

[0041]

[0042] Where τ represents the load data of the robotic arm joints, W F χ is the observation matrix of the joint parameters of the robotic arm. F For the joint parameters of the robotic arm to be identified.

[0043] Furthermore, steps S5 and S6 also include:

[0044] Collect the current group of robotic arm joint data, which includes the current robotic arm joint motion trajectory data and the current robotic arm joint load data;

[0045] The current robotic arm joint parameter observation matrix W is calculated based on the current robotic arm joint motion trajectory data. F And calculate the current observation matrix W of the robotic arm joint parameters. F The condition number cond(W F ):

[0046] cond(W F )=||W F ||·||W F -1 ||

[0047] Determine the current robotic arm joint parameter observation matrix W F The condition number cond(W F If the current robotic arm joint parameter observation matrix W is less than a preset value, then... F The current joint load data τ of the robotic arm is substituted into the overall observation equation of the robotic arm to calculate the current joint parameter χ of the robotic arm to be identified. F If not, discard the current group of robotic arm joint data and then obtain the next group of robotic arm joint data.

[0048] Furthermore, the current robotic arm joint parameter observation matrix W F The current joint load data τ of the robotic arm is substituted into the overall observation equation of the robotic arm to calculate the current joint parameter χ of the robotic arm to be identified. F In addition, it also includes:

[0049] The overall observation equation of the robotic arm is fitted using the least squares method to obtain the current joint parameter χ of the robotic arm to be identified. F The calculation formula is as follows:

[0050] χ F =(W F -1 W F )-1 W F τ

[0051] The current robotic arm joint parameter observation matrix W F Substituting the current robotic arm joint load data τ into the robotic arm joint parameter χ to be identified F The calculation formula is used to calculate the current joint parameter χ of the robotic arm to be identified. F .

[0052] Furthermore, it also includes:

[0053] Collect multiple sets of robotic arm joint data, each set of robotic arm joint data including robotic arm joint motion trajectory data and robotic arm joint load data;

[0054] The condition number cond(W) was selected from multiple sets of robotic arm joint data. F All groups of robotic arm joint data that are less than a preset value are considered, and the current group of robotic arm joint data is taken as the first group of robotic arm joint data among all groups of robotic arm joint data; wherein, the robotic arm joint parameter χ to be identified corresponds to the first group of robotic arm joint data. F,1 It is based on the current joint parameters χ of the robotic arm to be identified. F The calculation formula is obtained; the robot joint parameters to be identified corresponding to the other groups of robot joint data besides the first group of robot joint data are obtained according to the weighted iterative formula.

[0055] The robot arm joint parameters χ corresponding to the first set of robot arm joint data to be identified F,1 Substitute these values ​​into the weighted iterative formula to calculate the joint parameter χ of the robotic arm to be identified corresponding to the second set of robotic arm joint data. F,2 The second set of robotic arm joint data corresponds to the robotic arm joint parameters χ to be identified. F,2 Substituting these values ​​into the weighted iterative formula, the joint parameter χ of the robotic arm to be identified corresponding to the third set of robotic arm joint data is calculated. F,3 The joint parameters χ of the robotic arm to be identified corresponding to the third set of robotic arm joint data. F,3 Substituting these values ​​into the weighted iterative formula, the joint parameter χ of the robotic arm to be identified corresponding to the fourth set of robotic arm joint data is calculated. F,4 Similarly, the joint parameters χ of the robotic arm to be identified corresponding to the nth set of robotic arm joint data are... F,n Substitute these parameters into the weighted iterative formula to calculate the joint parameter χ of the robotic arm corresponding to the (n+1)th set of robotic arm joint data. F,n+1 The weighted iteration formula is as follows:

[0056]

[0057] Where, χ F,n+1 χ represents the joint parameters of the robotic arm to be identified corresponding to the (n+1)th group of robotic arm joint data. F,n W represents the joint parameters of the robotic arm to be identified corresponding to the nth set of robotic arm joint data. F,n+1 Let τ be the observation matrix of the robot arm joint parameters corresponding to the (n+1)th group of robot arm joint data. n+1 For the (n+1)th group of robotic arm joint data, τ represents the load data of the robotic arm joints. n This refers to the load data of the robotic arm joints in the nth group of robotic arm joint data.

[0058] Furthermore, the real-time acquisition of robotic arm joint motion trajectory data and robotic arm joint load data also includes:

[0059] The robotic arm's joint motion trajectory data is collected in real time by position sensors at each joint, and the robotic arm's joint load data is collected in real time by force sensors at each joint.

[0060] The present invention provides an online parameter identification method for heavy-duty robotic arms based on digital twins, which has the following advantages:

[0061] Beneficial effects:

[0062] (1) This invention does not require independent experimental system design and excitation trajectory design. It only needs to read the data of the current working of the robotic arm, which simplifies the identification conditions.

[0063] (2) The present invention can detect the parameter data of the robotic arm online in real time and can capture the changes in the parameters of the robotic arm caused by the work in real time;

[0064] (3) The identification method proposed in this invention has good versatility and can be applied to robotic arms with any number of axes and any configuration of joint force sensors. Attached Figure Description

[0065] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the following detailed description to explain the invention, but do not constitute a limitation thereof.

[0066] Figure 1 This is a framework diagram of a digital twin system in the existing technology.

[0067] Figure 2 The flowchart shows the online parameter identification method for heavy-duty robotic arms based on digital twins provided by this invention.

[0068] Figure 3 The flowchart shows the calculation process of the overall observation equation for the robotic arm provided by this invention.

[0069] Figure 4 The flowchart for calculating the joint parameters of the robotic arm to be identified is provided by the present invention.

[0070] Figure 5 This is a flowchart for weighted iteration of all selected data groups provided by the present invention. Detailed Implementation

[0071] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a method for online parameter identification of a heavy-duty robotic arm based on digital twins proposed according to the present invention. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the protection scope of the present invention.

[0072] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of the invention described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0073] This embodiment provides a method for online identification of heavy-duty robotic arm parameters based on digital twins, such as... Figure 2 As shown, the online parameter identification method for heavy-duty robotic arms based on digital twins includes the following steps:

[0074] Step S1: Derive the robotic arm dynamics formula according to the Newton-Euler recursive method to construct the robotic arm joint inertial equation, and linearize the robotic arm joint inertial equation to separate the robotic arm joint inertial parameter observation matrix and the robotic arm joint inertial parameters to be identified.

[0075] It should be noted that the Newton-Euler recursive method includes two aspects:

[0076] The first aspect: by using the velocity, angular velocity, acceleration, and angular acceleration of the robotic arm joints themselves, the velocity, angular velocity, acceleration, and angular acceleration of each axis of the robotic arm relative to the world coordinate system, as well as the velocity, angular velocity, acceleration, and angular acceleration of each joint of the robotic arm relative to the world coordinate system, are calculated. This process is a recursion from the base of the robotic arm to the end effector of the robotic arm.

[0077] The second aspect is that the load at the end of the robotic arm is used to progressively increase the load on each joint of the robotic arm based on the movement of each axis and joint. This process is a progressive increase from the end of the robotic arm to the base of the robotic arm.

[0078] Specifically, such as Figure 3 As shown, the process of deriving the robotic arm dynamics formulas using the Newton-Euler recursive method to construct the robotic arm joint inertia equations also includes:

[0079] The recursive process of the robotic arm dynamics formula is as follows:

[0080]

[0081] Where, ω or Represents the angular velocity vector. R represents the angular acceleration vector, R represents the rotation matrix, and v represents the velocity vector. The vector represents the acceleration vector, r is the relative position vector between the two joints, z is the direction vector of the z-axis of the robot arm coordinate system, f is the joint load force vector, μ is the joint load torque vector, F is the resultant force on the joint, and I is the joint moment of inertia. The subscript indicates the joint or axis number being described, and the superscript indicates the robot arm coordinate system number. The robot arm coordinate system is created sequentially from the base to the end effector using the DH method, and the homogeneous transformation matrix of each coordinate system is calculated. From this, the rotation matrix and the Z-axis direction vector are extracted from the homogeneous transformation matrix.

[0082] The joint inertia equation of the robotic arm is constructed based on the recursive process of the robotic arm dynamics formula.

[0083] Specifically, such as Figure 3 As shown, the linearization of the robotic arm joint inertia equations to separate the robotic arm joint inertia parameter observation matrix and the robotic arm joint inertia parameters to be identified further includes:

[0084] When the robotic arm is a six-axis robotic arm, the equations of inertia for the robotic arm joints are arranged as follows:

[0085]

[0086] Where, τ links This represents the inertial load data of the robotic arm joints, where Z is the zero matrix. For the unprocessed robotic arm joint inertial parameters, This is the unprocessed observation matrix of the robotic arm joint inertial parameters; wherein, the robotic arm joint motion trajectory data is input into the robotic arm joint inertial equation to calculate...

[0087] The minimum inertial parameters of the robotic arm are obtained based on the DH parameters and the joint type of the robotic arm. The determination of these minimum inertial parameters is based on Huo Wei's "Robot Dynamics and Control". These minimum inertial parameters are then used as the joint inertial parameters χ of the robotic arm to be identified. z The inertial parameters χ of the joints of the robotic arm to be identified z The columns are vectors, where the unprocessed robotic arm joint inertial parameters are... The inertial parameter χ of the joint of the robotic arm to be identified z The conversion relationship between them is as follows:

[0088]

[0089] Wherein, P is the unprocessed robotic arm joint inertial parameter. The inertial parameter χ of the joint of the robotic arm to be identified z The transformation matrix between the two parameters; this transformation relationship characterizes the linear relationship between the two parameters.

[0090] According to the transformation formula Transforming the joint inertia equations of the robotic arm, we obtain:

[0091]

[0092] Among them, W z This is the processed observation matrix of the robotic arm joint inertial parameters.

[0093] It should be noted that the minimum inertial parameters of this robotic arm can be obtained through matrix decomposition operations such as singular value decomposition or QR decomposition. Taking the UR5 robotic arm as an example, the joint types of the first to seventh joints are R3, R1, R1, R1, R1, R1. The minimum inertial parameter set of the link with joint type R3 includes only one parameter, and the minimum inertial parameter set of the link with joint type R1 includes 7 parameters. Therefore, the minimum inertial parameter set includes a total of 36 inertial parameters, which are listed as the joint inertial parameters χ of the robotic arm to be identified. z .

[0094] Step S2: Obtain the joint friction equation of the robotic arm and linearize the joint friction equation of the robotic arm to separate the observation matrix of joint friction parameters of the robotic arm and the joint friction parameters of the robotic arm to be identified;

[0095] Specifically, such as Figure 3 As shown, the step of obtaining the joint friction equation of the robotic arm and linearizing the joint friction equation to separate the observation matrix of the joint friction parameters of the robotic arm and the joint friction parameters to be identified of the robotic arm further includes:

[0096] The friction force equations for each joint of the robotic arm are obtained, and the friction force equations for each joint of the robotic arm are as follows:

[0097]

[0098] Where, τ fric,i Let be the frictional load data for the i-th joint of the robotic arm. Let μ be the velocity or angular velocity of the i-th joint of the robotic arm. i , λ i γ i These are the frictional force parameters to be identified for the i-th joint of the robotic arm;

[0099] The friction force equations for each joint of the robotic arm are linearized and expressed as vectors for each joint:

[0100]

[0101] The vectors of each joint of the robotic arm are arranged to obtain the joint friction force equation of the robotic arm, wherein the joint friction force equation of the robotic arm is as follows:

[0102]

[0103] Where, τ fric For the joint friction load data of the robotic arm, χ f To identify the frictional parameters of the robotic arm joints, W f This is the observation matrix for the joint friction parameters of the robotic arm; wherein, the motion trajectory data of the robotic arm joints is input into the joint friction equation of the robotic arm to calculate W. f .

[0104] Step S3: Merge the observation matrix of the inertial parameters of the robotic arm joint and the observation matrix of the frictional force of the robotic arm joint into a robotic arm joint parameter observation matrix, and at the same time merge the inertial parameters of the robotic arm joint to be identified and the frictional force parameters of the robotic arm joint to be identified into the joint parameters of the robotic arm to be identified.

[0105] Step S4: Obtain the overall observation equation of the robotic arm based on the observation matrix of the robotic arm joint parameters and the joint parameters of the robotic arm to be identified;

[0106] Preferably, the step of obtaining the overall observation equation of the robotic arm based on the observation matrix of the robotic arm joint parameters and the joint parameters of the robotic arm to be identified further includes:

[0107] The joint inertia equation and the joint friction equation of the robotic arm are combined to obtain the overall observation equation of the robotic arm, which is as follows:

[0108]

[0109] Where τ represents the load data of the robotic arm joints, W F χ is the observation matrix of the joint parameters of the robotic arm. F For the joint parameters of the robotic arm to be identified.

[0110] Step S5: Collect robot arm joint data in real time under any working condition, and filter the collected robot arm joint data to obtain filtered robot arm joint data; wherein, the robot arm joint data includes robot arm joint motion trajectory data and robot arm joint load data, and the robot arm joint load data includes robot arm joint inertial load data and robot arm joint friction load data.

[0111] It should be noted that the data source is the robotic arm in operation. Sensors acquire its operational data in real time and then transmit it to the digital twin system. Specifically, the robotic arm joint motion trajectory data is collected in real time through position sensors of each joint or the input trajectory of the robotic arm, and the robotic arm joint load data is collected in real time through force sensors of each joint. The robotic arm joint motion trajectory data includes the velocity, angular velocity, acceleration, and angular acceleration of each joint.

[0112] Step S6: Substitute the filtered robotic arm joint data into the overall observation equation of the robotic arm to calculate the joint parameters of the robotic arm to be identified.

[0113] Preferably, such as Figure 4 As shown, steps S5 and S6 further include:

[0114] Collect the current group of robotic arm joint data, which includes the current robotic arm joint motion trajectory data and the current robotic arm joint load data;

[0115] The current robotic arm joint parameter observation matrix W is calculated based on the current robotic arm joint motion trajectory data. F And calculate the current observation matrix W of the robotic arm joint parameters. F The condition number cond(W F ):

[0116] cond(W F )=||W F ||·||W F -1 ||

[0117] While there is no universally accepted standard for measuring matrix norm, experiments have shown that the 1-norm is the best approach. When W... F When it is not a square matrix, its inverse matrix can be a generalized inverse matrix.

[0118] Determine the current robotic arm joint parameter observation matrix W F The condition number cond(W F If the current robotic arm joint parameter observation matrix W is less than a preset value, then... F The current joint load data τ of the robotic arm is substituted into the overall observation equation of the robotic arm to calculate the current joint parameter χ of the robotic arm to be identified. F If not, discard the current group of robotic arm joint data and then obtain the next group of robotic arm joint data.

[0119] It should be noted that this preset value is not fixed and depends on the desired recognition efficiency and accuracy. The more stringent the condition number standard, the higher the recognition accuracy but the lower the recognition efficiency. Experiments have shown that a preset value between 5 and 10 is optimal.

[0120] Specifically, the current robotic arm joint parameter observation matrix W F The current joint load data τ of the robotic arm is substituted into the overall observation equation of the robotic arm to calculate the current joint parameter χ of the robotic arm to be identified. F In addition, it also includes:

[0121] The overall observation equation of the robotic arm is fitted using the least squares method to obtain the current joint parameter χ of the robotic arm to be identified. F The calculation formula is as follows:

[0122] χ F =(W F -1 W F ) -1 W F τ

[0123] The current robotic arm joint parameter observation matrix W F Substituting the current robotic arm joint load data τ into the robotic arm joint parameter χ to be identified F The calculation formula is used to calculate the current joint parameter χ of the robotic arm to be identified. F .

[0124] It should be noted that the least squares method can also be replaced by linear ARX model methods, model training methods, etc., and this invention does not limit the method.

[0125] Preferably, such as Figure 5 As shown, it also includes:

[0126] Collect multiple sets of robotic arm joint data under any working condition. Each set of robotic arm joint data includes robotic arm joint motion trajectory data and robotic arm joint load data.

[0127] It should be noted that the data range used for this parameter identification should be defined first. The more data, the more accurate it will be. However, if the time span is too long, it will be impossible to capture parameter changes. It is recommended to do it in two ways: one is online detection during the working process, such as once per minute, and the other is offline detection throughout the entire working process.

[0128] The condition number cond(W) was selected from multiple sets of robotic arm joint data. F All groups of robotic arm joint data that are less than a preset value are considered, and the current group of robotic arm joint data is taken as the first group of robotic arm joint data among all groups of robotic arm joint data; wherein, the robotic arm joint parameter χ to be identified corresponds to the first group of robotic arm joint data. F,1 It is based on the current joint parameters χ of the robotic arm to be identified. F The calculation formula is obtained; the robot joint parameters to be identified corresponding to the other groups of robot joint data besides the first group of robot joint data are obtained according to the weighted iterative formula.

[0129] The robot arm joint parameters χ corresponding to the first set of robot arm joint data to be identified F,1 Substitute these values ​​into the weighted iterative formula to calculate the joint parameter χ of the robotic arm to be identified corresponding to the second set of robotic arm joint data. F,2 The second set of robotic arm joint data corresponds to the robotic arm joint parameters χ to be identified. F,2 Substituting these values ​​into the weighted iterative formula, the joint parameter χ of the robotic arm to be identified corresponding to the third set of robotic arm joint data is calculated. F,3 The joint parameters χ of the robotic arm to be identified corresponding to the third set of robotic arm joint data. F,3 Substituting these values ​​into the weighted iterative formula, the joint parameter χ of the robotic arm to be identified corresponding to the fourth set of robotic arm joint data is calculated. F,4 Similarly, the joint parameters χ of the robotic arm to be identified corresponding to the nth set of robotic arm joint data are... F,n Substitute these parameters into the weighted iterative formula to calculate the joint parameter χ of the robotic arm corresponding to the (n+1)th set of robotic arm joint data. F,n+1 The weighted iteration formula is as follows:

[0130]

[0131] Where, χ F,n+1 χ represents the joint parameters of the robotic arm to be identified corresponding to the (n+1)th group of robotic arm joint data.F,n W represents the joint parameters of the robotic arm to be identified corresponding to the nth set of robotic arm joint data. F,n+1 Let τ be the observation matrix of the robot arm joint parameters corresponding to the (n+1)th group of robot arm joint data. n+1 For the (n+1)th group of robotic arm joint data, τ represents the load data of the robotic arm joints. n This refers to the load data of the robotic arm joints in the nth group of robotic arm joint data.

[0132] It should be noted that a weighted fitting method is used for the data in all groups except the first group, and this is done stepwise. The advantage of this method is that it avoids reprocessing the previous data when processing newly received data groups. The detailed process is as follows:

[0133] Substitute the first set of data from all sets into the current joint parameter χ of the robotic arm to be identified. F The calculation formula for χ is: F =(W F -1 W F ) -1 W F τ can be used to calculate the joint parameters χ of the robotic arm to be identified corresponding to the first set of data. F,1 ;

[0134] Having fitted n sets of data, for the next set of data τ received... n+1 W F,n+1 Perform weighted iteration:

[0135]

[0136] The weighted iterative formula can be used to calculate a weighted average of newly detected data and previous data in a single detection, enabling rapid data processing. This means that data processing is performed simultaneously with data entry, rather than after all data has been entered, thereby maximizing the use of computer performance and shortening response time.

[0137] The present invention provides an online parameter identification method for heavy-duty robotic arms based on digital twins, which (1) constructs a data transmission model related to the digital twin system, including data transmission methods and data processing methods within the system; (2) uses joint force sensors and displacement sensors of the robotic arm to obtain the dynamic data of the robotic arm, transmits the data to the digital twin system, establishes a dynamic model of the robotic arm within the digital twin system, calculates the dynamics of the robotic arm based on the acquired data under the dynamic model of the robotic arm, and finally realizes the parameter identification process.

[0138] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for online parameter identification of a heavy-duty robotic arm based on digital twins, characterized in that, Includes the following steps: Step S1: Derive the robotic arm dynamics formula according to the Newton-Euler recursive method to construct the robotic arm joint inertia equation, and linearize the robotic arm joint inertia equation to separate the robotic arm joint inertia parameter observation matrix and the robotic arm joint inertia parameters to be identified. Step S2: Obtain the joint friction equation of the robotic arm and linearize the joint friction equation of the robotic arm to separate the observation matrix of joint friction parameters of the robotic arm and the joint friction parameters of the robotic arm to be identified; Step S3: Merge the observation matrix of the inertial parameters of the robotic arm joint and the observation matrix of the frictional force of the robotic arm joint into a robotic arm joint parameter observation matrix, and at the same time merge the inertial parameters of the robotic arm joint to be identified and the frictional force parameters of the robotic arm joint to be identified into the joint parameters of the robotic arm to be identified. Step S4: Obtain the overall observation equation of the robotic arm based on the observation matrix of the robotic arm joint parameters and the joint parameters of the robotic arm to be identified; Step S5: Collect robotic arm joint data in real time and filter the collected robotic arm joint data to obtain filtered robotic arm joint data; wherein, the robotic arm joint data includes robotic arm joint motion trajectory data and robotic arm joint load data, and the robotic arm joint load data includes robotic arm joint inertial load data and robotic arm joint friction load data. Step S6: Substitute the filtered robotic arm joint data into the overall observation equation of the robotic arm to calculate the joint parameters of the robotic arm to be identified.

2. The method for online identification of heavy-duty robotic arm parameters based on digital twins according to claim 1, characterized in that, The process of deriving the robotic arm dynamics formulas using the Newton-Euler recursive method to construct the robotic arm joint inertia equations also includes: The recursive process of the robotic arm dynamics formula is as follows: Where, ω or Represents the angular velocity vector. R represents the angular acceleration vector, R represents the rotation matrix, and v represents the velocity vector. The vector represents the acceleration vector, r is the relative position vector between the two joints, z is the direction vector of the z-axis of the robot arm coordinate system, f is the joint load force vector, μ is the joint load torque vector, F is the resultant force on the joint, and I is the joint moment of inertia. The subscript indicates the joint or axis number being described, and the superscript indicates the robot arm coordinate system number. The robot arm coordinate system is created sequentially from the base to the end effector using the DH method, and the homogeneous transformation matrix of each coordinate system is calculated. From this, the rotation matrix and the Z-axis direction vector are extracted from the homogeneous transformation matrix. The joint inertia equation of the robotic arm is constructed based on the recursive process of the robotic arm dynamics formula.

3. The method for online identification of heavy-duty robotic arm parameters based on digital twins according to claim 2, characterized in that, The Newton-Euler recursive method includes two aspects: The first aspect: by using the velocity, angular velocity, acceleration, and angular acceleration of the robotic arm joints themselves, the velocity, angular velocity, acceleration, and angular acceleration of each axis of the robotic arm relative to the world coordinate system, as well as the velocity, angular velocity, acceleration, and angular acceleration of each joint of the robotic arm relative to the world coordinate system, are calculated. This process is a recursion from the base of the robotic arm to the end effector of the robotic arm. The second aspect is that the load at the end of the robotic arm is used to progressively increase the load on each joint of the robotic arm based on the movement of each axis and joint. This process is a progressive increase from the end of the robotic arm to the base of the robotic arm.

4. The method for online identification of heavy-duty robotic arm parameters based on digital twins according to claim 2, characterized in that, The process of linearizing the robotic arm joint inertia equations to separate the robotic arm joint inertia parameter observation matrix and the robotic arm joint inertia parameters to be identified also includes: When the robotic arm is a six-axis robotic arm, the equations of inertia for the robotic arm joints are arranged as follows: Where, τ links This represents the inertial load data of the robotic arm joints, where Z is the zero matrix. For the unprocessed robotic arm joint inertial parameters, This is the unprocessed observation matrix of the robotic arm joint inertial parameters; wherein, the robotic arm joint motion trajectory data is input into the robotic arm joint inertial equation to calculate... The minimum inertia parameter of the robotic arm is obtained based on the DH parameters and the joint type of the robotic arm, and this minimum inertia parameter is used as the inertia parameter χ of the joint of the robotic arm to be identified. z The unprocessed robotic arm joint inertial parameters The inertial parameter χ of the joint of the robotic arm to be identified z The conversion relationship between them is as follows: Wherein, P is the unprocessed robotic arm joint inertial parameter. The inertial parameter χ of the joint of the robotic arm to be identified z The transformation matrix between them; According to the transformation formula Transforming the joint inertia equations of the robotic arm, we obtain: Among them, W z This is the processed observation matrix of the robotic arm joint inertial parameters.

5. The method for online identification of heavy-duty robotic arm parameters based on digital twins according to claim 4, characterized in that, The process of obtaining the joint friction equation of the robotic arm and linearizing the joint friction equation to separate the observation matrix of the joint friction parameters of the robotic arm and the joint friction parameters to be identified also includes: The friction force equations for each joint of the robotic arm are obtained, and the friction force equations for each joint of the robotic arm are as follows: Where, τ fric,i Let be the frictional load data for the i-th joint of the robotic arm. Let μ be the velocity or angular velocity of the i-th joint of the robotic arm. i , λ i γ i These are the frictional force parameters to be identified for the i-th joint of the robotic arm; The friction force equations for each joint of the robotic arm are linearized and expressed as vectors for each joint: The vectors of each joint of the robotic arm are arranged to obtain the joint friction force equation of the robotic arm, wherein the joint friction force equation of the robotic arm is as follows: Where, τ fric For the joint friction load data of the robotic arm, χ f To identify the frictional parameters of the robotic arm joints, W f This is the observation matrix for the joint friction parameters of the robotic arm; wherein, the motion trajectory data of the robotic arm joints is input into the joint friction equation of the robotic arm to calculate W. f .

6. The method for online identification of heavy-duty robotic arm parameters based on digital twins according to claim 5, characterized in that, The process of obtaining the overall observation equation for the robotic arm based on the observation matrix of the robotic arm joint parameters and the joint parameters of the robotic arm to be identified also includes: The joint inertia equation and the joint friction equation of the robotic arm are combined to obtain the overall observation equation of the robotic arm, which is as follows: Where τ represents the load data of the robotic arm joints, W F χ is the observation matrix of the joint parameters of the robotic arm. F For the joint parameters of the robotic arm to be identified.

7. The method for online identification of heavy-duty robotic arm parameters based on digital twins according to claim 6, characterized in that, Steps S5 and S6 further include: Collect the current group of robotic arm joint data, which includes the current robotic arm joint motion trajectory data and the current robotic arm joint load data; The current robotic arm joint parameter observation matrix W is calculated based on the current robotic arm joint motion trajectory data. F And calculate the current observation matrix W of the robotic arm joint parameters. F The condition number cond(W F ): cond(W F )=||W F ||·||W F -1 || Determine the current robotic arm joint parameter observation matrix W F The condition number cond(W F If the current robotic arm joint parameter observation matrix W is less than a preset value, then... F The current joint load data τ of the robotic arm is substituted into the overall observation equation of the robotic arm to calculate the current joint parameter χ of the robotic arm to be identified. F If not, discard the current group of robotic arm joint data and then obtain the next group of robotic arm joint data.

8. The method for online identification of heavy-duty robotic arm parameters based on digital twins according to claim 7, characterized in that, The current robotic arm joint parameter observation matrix W F The current joint load data τ of the robotic arm is substituted into the overall observation equation of the robotic arm to calculate the current joint parameter χ of the robotic arm to be identified. F In addition, it also includes: The overall observation equation of the robotic arm is fitted using the least squares method to obtain the current joint parameter χ of the robotic arm to be identified. F The calculation formula is as follows: x F =(W F -1 W F ) -1 W F t The current robotic arm joint parameter observation matrix W F Substituting the current robotic arm joint load data τ into the robotic arm joint parameter χ to be identified F The calculation formula is used to calculate the current joint parameter χ of the robotic arm to be identified. F .

9. The method for online identification of heavy-duty robotic arm parameters based on digital twins according to claim 8, characterized in that, Also includes: Collect multiple sets of robotic arm joint data, each set of robotic arm joint data including robotic arm joint motion trajectory data and robotic arm joint load data; The condition number cond(W) was selected from multiple sets of robotic arm joint data. F All groups of robotic arm joint data that are less than a preset value are considered, and the current group of robotic arm joint data is taken as the first group of robotic arm joint data among all groups of robotic arm joint data; wherein, the robotic arm joint parameter χ to be identified corresponds to the first group of robotic arm joint data. F，1 It is based on the current joint parameters χ of the robotic arm to be identified. F The calculation formula is obtained; the robot joint parameters to be identified corresponding to the other groups of robot joint data besides the first group of robot joint data are obtained according to the weighted iterative formula. The joint parameter χ of the robotic arm to be identified corresponding to the first set of robotic arm joint data F，1 Substituting these values ​​into the weighted iterative formula, the joint parameter χ of the robotic arm to be identified corresponding to the second set of robotic arm joint data is calculated. F，2 The second set of robotic arm joint data corresponds to the robotic arm joint parameters χ to be identified. F，2 Substituting these values ​​into the weighted iterative formula, the joint parameter χ of the robotic arm to be identified corresponding to the third set of robotic arm joint data is calculated. F，3 The joint parameters χ of the robotic arm to be identified corresponding to the third set of robotic arm joint data. F，3 Substituting these values ​​into the weighted iterative formula, the joint parameter χ of the robotic arm to be identified corresponding to the fourth set of robotic arm joint data is calculated. F，4 Similarly, the joint parameters χ of the robotic arm to be identified corresponding to the nth set of robotic arm joint data are... F，n Substitute these parameters into the weighted iterative formula to calculate the joint parameter χ of the robotic arm corresponding to the (n+1)th group of robotic arm joint data. F，n+1 The weighted iteration formula is as follows: Where, χ F,n+1 χ represents the joint parameters of the robotic arm to be identified corresponding to the (n+1)th group of robotic arm joint data. F，n W represents the joint parameters of the robotic arm to be identified corresponding to the nth set of robotic arm joint data. F，n+1 Let τ be the observation matrix of the robot arm joint parameters corresponding to the (n+1)th group of robot arm joint data. n+1 For the (n+1)th group of robotic arm joint data, τ represents the load data of the robotic arm joints. n This refers to the load data of the robotic arm joints in the nth group of robotic arm joint data.

10. The method for online identification of heavy-duty robotic arm parameters based on digital twins according to claim 1, characterized in that, The real-time acquisition of robotic arm joint motion trajectory data and robotic arm joint load data also includes: The robotic arm's joint motion trajectory data is collected in real time by position sensors at each joint, and the robotic arm's joint load data is collected in real time by force sensors at each joint.

Images