CN-122008204-A - Dynamic modeling and friction parameter identification method for hybrid robot

CN122008204ACN 122008204 ACN122008204 ACN 122008204ACN-122008204-A

Abstract

The invention provides a dynamic modeling and friction parameter identification method of a series-parallel robot, which comprises the following steps of establishing a dynamic model considering active-passive joint friction for the series-parallel robot by utilizing a Newton-Euler method and a spiral theory, carrying out global sensitivity analysis on friction parameters of the series-parallel robot by adopting a sum function variance and covariance decomposition method, planning identification and verification circular tracks of the series-parallel robot according to a five-time polynomial, collecting driving torque of a servo motor of the robot in a motion process, and carrying out friction parameter identification on the series-parallel robot by utilizing a Gaussian quantum particle swarm algorithm. The method has the advantages that the built model can accurately predict the driving moment under different configurations, and a reliable dynamic basis is provided for track planning and real-time control of the robot.

Inventors

MA YUE
AI CHANG
PENG DUN
PENG HONGGUI
CHEN LONG
LI BIN
WU QIRONG
LIU QI
Hao Tianze
LI PENGCHAO
SUN WEILONG
LI CHENGRUI

Assignees

天津理工大学

Dates

Publication Date: 20260512
Application Date: 20260130

Claims (9)

1. The method for dynamic modeling and friction parameter identification of the hybrid robot is characterized by comprising the following steps of: S1, establishing a dynamic model considering active and passive joint friction for a hybrid robot by utilizing a Newton-Euler method and a spiral theory; s2, carrying out global sensitivity analysis on friction parameters of the series-parallel robot by adopting a sum function variance and covariance decomposition method; S3, planning the identification and verification circular track of the hybrid robot according to the quintic polynomial; s4, operating the series-parallel robot to move according to a set track, and collecting driving torque of a servo motor in the movement process of the robot; S5, carrying out friction parameter identification on the hybrid robot by utilizing a Gaussian quantum particle swarm algorithm.
2. The method for dynamic modeling and friction parameter identification of a hybrid robot according to claim 1, wherein S1 comprises the following sub-steps: s11, according to the inverse kinematics model, passing through the tail end reference point The position, the speed and the acceleration of each active joint variable of the hybrid robot are obtained 、 Speed amplitude 、 Acceleration amplitude 、 ; S12, numbering all components of the series-parallel robot needing to consider inertia, wherein the driving arm is split into a sleeve, a screw rod, a motor assembly and a push rod assembly, and the sleeve, the screw rod, the motor assembly and the push rod assembly are arranged on a branched chain The device is characterized in that a sleeve, a screw rod and a motor assembly are arranged in the device, a push rod assembly is a component 1, a rotating bracket is arranged in a branched chain 4, a component 0, a universal ring assembly is a component 1, a driven support arm is a component 2, a movable platform is a component 3, a first swinging head is a component 4, a second swinging head is a component 5, and a cutter is a component 6; S13, obtaining all components and points according to the speeds and accelerations of all joints Instantaneous coincident point Velocity/acceleration helix of (2) / , In (1), if , If (1) , ; S14, calculating the point of all the components by the following formula Resultant force screw of inertial force/moment and gravity force : ; In the formula, Indicating the point of the member Is a spatial inertial matrix of (a); Representing a spiral Cross product operator of (a); indicating the point of the member Is a gravity screw of (2); S15, according to Newton-Euler method, all the components are regarded as free bodies to be stressed, stress analysis is carried out on the free bodies, any one component is acted by gravity, inertia force/moment, constraint force/moment and friction force/moment of kinematic pairs at two ends, if the kinematic pair is an active pair, the driving force/moment is also included, and the reference point of the kinematic pair at two ends of any one component is set as The constraint resultant force spiral is written by the following formula , wherein, Representing component reference points Is provided with a constraining helix of (a), Representing component reference points Is a constrained helix of (2): = ; In the formula, Representation points Unit vectors of three orthogonal directions; Representation points To the point Is a direction vector of (2); an amplitude matrix representing the constraint helix in three orthogonal directions, expressed as: ; In the formula, 、、 And Respectively expressed at reference points Along the direction A restraining force amplitude, a friction force amplitude, a restraining torque amplitude, and a friction torque amplitude, wherein, if Representing the reference point of the active pair and Representing the axis vector of the active pair, then the primary mobile pair Indicating driving force, driving revolute pair Representing the driving torque; s16, modeling the friction force/moment of the kinematic pair by adopting a Stribeck model, wherein the expression is as follows: ; In the formula, The expression friction force/moment is used, 、、 And Respectively representing the coulomb friction coefficient, the maximum static friction coefficient, the viscous friction coefficient and the Stribeck speed; Representing the radius of the friction circle of the kinematic pair, wherein the kinematic pair ; Representing the magnitude of the restraining force on the vertical plane of the axis of the kinematic pair; sign represents a sign function; S17, writing all the component about points by the following formula Newton-euler equation of (v): ; S18, finishing Newton-Euler equations of all components, wherein the Newton-Euler equations comprise 78 equations, 80 unknown constraint forces/moments, and combining 2 equations of over-constraint deformation-coordination conditions to form 80 equations, and 80 unknown linear equation sets: ; In the formula, The matrix of coefficients is represented and, The vector of the constants is represented as, Representing an unknown parameter vector comprising joint constraint force/moment and driving force/moment; S19, obtaining according to solving the linear equation set And extract joint driving force of 1T2R parallel mechanism from the joint driving force Joint driving moment of first and second swinging heads and cutter ; S110, driving the joint by the following formula The conversion into driving torque: ; In the formula, Indicating the lead of the ball screw, Representing the friction torque at the screw-nut.
3. The method for dynamic modeling and friction parameter identification of the hybrid robot according to claim 2, wherein the step S19 comprises the following sub-steps: S191, under the condition of neglecting joint friction influence, solving a linear equation set of S18 to obtain constraint force/moment of each joint; s192, bringing the obtained joint constraint force into the Stribeck friction model of S16 to obtain the friction force/moment of each joint; S193, considering the influence of neglecting joint friction on the solving precision of the joint constraint force in S191, taking the joint friction force/moment obtained in S192 into stress analysis, and obtaining the joint constraint force after primary correction by means of the linear equation set of S18; S194, repeating S192 and S193 to obtain the joint driving and restraining force after the secondary correction.
4. A method for dynamic modeling and friction parameter identification of a hybrid robot, wherein the hybrid robot comprises 6 active joints and 12 passive joints, the active joints are directly incorporated into an identification process because of the dominant role in friction effect, and all the passive joints are positioned in a 1T2R parallel mechanism, and the method is characterized in that the application of S2 of any one of claims 1-3 comprises the following sub-steps: S21, dividing friction parameters of all passive joints into four types, namely coulomb friction, viscous friction, static friction and Stribeck speed; s22, taking coulomb friction parameters as an example, constructing 3 driving forces of the 1T2R parallel mechanism by the following formula 3 Sets of summation functions of (c): , ; In the formula, A combination vector representing 12 of the coulomb friction parameters; S23, will 、 Represented by a high-dimensional model as the sum of a series of low-dimensional functions, And (3) with The variance of (c) can be expressed as: ; s24, independently sampling the coulomb friction parameter twice in the sampling space by means of Latin super-vertical sampling method to generate Sample matrix of row 12 columns And (3) with Number of samples Set to 15000, then exchange And (3) with Is the first of (2) Column, construction matrix Here, the number of the first and second electrodes, For representation Is the first of (2) Column replacement First, the The column, and thus, 、 The variance of (c) and its subfunction variance can be expressed as: , ; ; S25, the total fluctuation of the system is calculated The definition is as follows: ; S26, coulomb friction parameter Separately induced partial fluctuations and other coulomb friction parameters The sum of all the partial fluctuations caused by other coulomb friction parameters, except the partial fluctuations of the interaction, is defined as: ; In the formula, Representation of Except for inclusion of In addition to the sub-functions of (c), the sum of the variances of the remaining sub-functions, Representation of Except for inclusion of In addition to the sub-functions of (2), the sum of the variances of the remaining sub-functions; s27, defining Coulomb friction parameters Global sensitivity index of (2) The method comprises the following steps: ; s28, repeating S22 to S27 to obtain global sensitivity indexes of viscous friction, static friction and Stribeck speed ; S29 according to four types of friction And selecting the high-sensitivity friction parameters of the passive joint, incorporating the friction parameters, and ignoring the influence of the low-sensitivity friction parameters.
5. The method for dynamic modeling and friction parameter identification of a hybrid robot according to claim 1, wherein S3 comprises the following sub-steps: s31, setting the speed and acceleration of the robot at the starting and stopping time to be 0, and starting the angle Angle of termination Start time is The termination time is Dots (dot) Connection line with circle center of circular track and coordinate system Included angle of y-axis of (2) The change rule of the fifth order polynomial is satisfied: ; In the formula, Term coefficients of 0 to 5 degree for the corresponding fifth degree polynomial; s32, obtaining a point from S31 Is a speed helix of (2) For a pair of Deriving to obtain acceleration helix 。
6. The method for dynamic modeling and friction parameter identification of a hybrid robot according to claim 1, wherein S4 comprises the following steps: s41, utilizing inverse kinematics of the series-parallel robot, according to the point The position, the speed and the acceleration of each active joint are calculated; S42, compiling the acquired active joint kinematics information into a G code, and then inputting the G code into Power PMAC IDE software to drive the hybrid robot to move according to the identification and verification circular track; s43, synchronously collecting driving current signals of each servo motor of two tracks in the motion process of the series-parallel robot ; S44, converting the driving current signal into driving torque by the following formula: ; In the formula, For the reduction ratio of the speed reducer connected with the servo motor, Is the torque constant of the servo motor, Is the servo motor current.
7. The method for dynamic modeling and friction parameter identification of a hybrid robot according to claim 1, wherein S5 comprises the following sub-steps: S51, determining the number of particles Initializing friction parameter groups in a friction parameter value space Wherein Representing friction parameter combinations to be identified; s52, taking the sum of the mean square error of the theoretical driving moment of each active joint S19 and S110 and the acquisition driving moment of S44 as a fitness function, and setting the minimum fitness function as an optimizing target, wherein the expression is as follows: ; In the formula, Representing the number of sampling points in the parameter identification, 、 Respectively representing a theoretical driving moment value and an acquisition driving moment value of an ith active joint at a jth sampling point; s53, calculating the fit value of each particle, before each particle is compared Fil values of the generation and the generation, thereby updating the local optimum friction parameter combination of the particle individuals Comparing the fit values of all the particles in the previous t generation and the current generation, thereby updating the globally optimal friction parameter combination of the particles ; S54, calculating variables by the following formula : ; In the formula, And (3) with Is a Gaussian mutation operator and obeys Gaussian distribution with the mean value of 0 and the variance of 1; S55, contraction-expansion factor Dynamic adjustment is realized through a linear decremental strategy, and the expression is as follows: ; In the formula, And Respectively representing the initial value and the final value of the inertia weight, The number of iterations of the maximum is indicated, Representing the current iteration number; S56, calculating the local optimal friction parameter combination of all particles by the following formula Arithmetic mean vector of (a) : ; S57, updating all particle states by the following formula: ; In the formula, Representing a random number; s58, jumping to S53 until the maximum iteration number is reached, and performing the last generation And the corresponding globally optimal friction parameter combination is used as an identification result of the friction parameters.
8. The method for identifying dynamic modeling and friction parameters of a hybrid robot according to claim 1, further comprising S6, verifying and evaluating the dynamic model through experiments according to the identified parameters, and correcting the dynamic model of the hybrid robot.
9. The method for dynamic modeling and friction parameter identification of a hybrid robot according to claim 8, wherein S6 comprises the following sub-steps: s61, calculating the driving moment of each active joint by using the identified friction parameters, and drawing a comparison chart of the theoretical driving moment and the acquired driving moment; s62, using the determination coefficient The kinetic model is evaluated, and the expression is as follows: ; In the formula, Represent the first The average value of the driving moment acquired by the active joints in N sampling points; s63, in order to further verify the generalization capability of the dynamic model and the friction parameter identification result, calculating the driving moment of each active joint under the verification track, drawing a comparison chart of the theoretical driving moment and the acquired driving moment, and using S62 The kinetic model was evaluated.

Description

Dynamic modeling and friction parameter identification method for hybrid robot Technical Field The invention relates to the field of industrial robot dynamics, in particular to a method for modeling dynamics and identifying friction parameters of a hybrid robot. Background The series-parallel robot combines the advantages of large working space of the series-parallel robot, high rigidity and high bearing capacity of the parallel robot, and has obvious application value in the high-end industrial fields of aerospace component processing, high-end equipment manufacturing and the like. The dynamic model is used as a basis for realizing high-precision control, the stress and motion relation of each part in the motion process of the robot is required to be accurately reflected, joint friction is used as one of key factors influencing the precision of the dynamic model, and the dynamic model has non-negligible effects on the pose error, the track tracking performance and the stability of the robot. At present, research on robot dynamics modeling has been advanced to a certain extent, but there is still a clear gap between the comprehensiveness of friction consideration and model accuracy, on the one hand, most of the existing dynamics modeling methods focus on friction parameter modeling of an active joint, often neglect friction influence of a passive joint, however, in a hybrid robot, the passive joint is used as a key link of motion transmission, friction of the passive joint can generate accumulated errors along with motion configuration changes, and particularly when the dynamic model moves at a high speed or in a complex track, the errors can obviously reduce the accuracy of the dynamic model, on the other hand, joint friction force is often regarded as a constant value in the traditional modeling process, or calculation is simplified only through an empirical formula, and dynamic influence of joint constraint force on friction is not considered, so that the dynamic model is difficult to accurately match with actual working conditions. In the aspect of friction parameter identification, the existing identification method has the outstanding problems that firstly systematic screening of friction parameters is lacking, all parameters in a model are generally identified uniformly, in practice, the sensitivity of part of friction parameters to driving torque is low, computational redundancy is increased due to blind inclusion identification, the coupling between an identification process and a dynamic model is insufficient, constraint force change under the practical configuration of a robot is not combined, the adaptability of the identified parameters under different motion working conditions is poor, and the identified parameters are difficult to be directly applied to a high-precision dynamic model. Disclosure of Invention In view of the above, the present invention is directed to a hybrid robot dynamic modeling and friction parameter identification method, so as to solve at least one of the above problems in the prior art. In order to achieve the above purpose, the technical scheme of the invention is realized as follows: a dynamic modeling and friction parameter identification method of a hybrid robot comprises the following steps: S1, establishing a dynamic model considering active and passive joint friction for a hybrid robot by utilizing a Newton-Euler method and a spiral theory; s2, carrying out global sensitivity analysis on friction parameters of the series-parallel robot by adopting a sum function variance and covariance decomposition method; S3, planning the identification and verification circular track of the hybrid robot according to the quintic polynomial; s4, operating the series-parallel robot to move according to a set track, and collecting driving torque of a servo motor in the movement process of the robot; S5, carrying out friction parameter identification on the hybrid robot by utilizing a Gaussian quantum particle swarm algorithm. Further, S1 comprises the following sub-steps: s11, according to the inverse kinematics model, passing through the tail end reference point The position, the speed and the acceleration of each active joint variable of the hybrid robot are obtained、Speed amplitude、Acceleration amplitude、; S12, numbering all components of the series-parallel robot with inertia to be considered, wherein the driving arm can be split into a sleeve, a screw rod and motor assembly (without considering the rotational inertia of the screw rod around the axis of the screw rod) and a push rod assembly, and the driving arm is branched in the following wayThe middle sleeve, the screw rod and the motor component are arranged as a component 1, and the push rod component is arranged as a component 2; the first swinging head, the second swinging head and the cutter are respectively a component 4, a component 5 and a component 6; S13, obtaining all components and points from the speeds and acc