Industrial robot kinetic parameter identification method based on adaptive mutation particle swarm optimization

Abstract

The invention provides an industrial robot kinetic parameter identification method based on an adaptive mutation particle swarm algorithm, and the method comprises the following steps: S1, building a kinetic model according to a DH parameter of a robot and a Newton-Euler kinetic equation, and directly determining a minimum parameter set according to the DH parameter and a joint type; s2, optimizing two mixed trajectories of Fourier series and polynomial in MATLAB (matrix laboratory); s3, driving the robot to run an excitation track, collecting data, and performing filtering calculation to obtain an observation matrix and a joint torque; s4, calculating a group of initial solutions through a least square method, and taking the initial solutions as a particle swarm optimization algorithm search range of adaptive variation in a certain range of the initial solutions; and S5, driving the robot to run the verification track, and processing the collected data to obtain an observation matrix of the verification track. According to the method, the mixed trajectory based on the Fourier series and the polynomial serves as the excitation trajectory, one-time overall identification is carried out through the self-adaptive variation particle swarm optimization algorithm, the identification result is accurate, and the predicted torque following performance is high.

Description

Technical Field

[0001] The present invention relates to the technical field related to industrial robots, and in particular to an industrial robot dynamics parameter identification method based on an adaptive mutation particle swarm algorithm. Background Art

[0002] The identification accuracy of robot dynamic parameters can improve control accuracy and is the basis of feedforward control. It is used in collision detection, drag teaching and other fields. The current difficulty in dynamic parameter identification is that accurate parameter values ​​cannot be obtained because the robot has many dynamic parameters and the coupling between joints is serious. In order to improve the accuracy of overall robot parameter identification, the present invention provides an industrial robot dynamic parameter identification method based on an adaptive mutation particle swarm algorithm. Summary of the invention

[0003] In view of the deficiencies in the prior art, the present invention provides an industrial robot dynamics parameter identification method based on an adaptive mutation particle swarm algorithm, which solves the problems mentioned in the background technology.

[0004] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0005] The method for identifying the dynamic parameters of industrial robots based on adaptive mutation particle swarm algorithm includes the following steps:

[0006] S1. Establish a dynamic model based on the robot's DH parameters and Newton-Euler dynamic equations, linearize the dynamic model, and directly determine the minimum parameter set based on the DH parameters and joint types;

[0007] S2. Optimize two hybrid trajectories of Fourier series and polynomials in MATLAB, one as the excitation trajectory and the other as the verification trajectory;

[0008] S3, based on step S2, drive the robot to run step S2 to optimize the excitation trajectory, collect data and filter and calculate to obtain the observation matrix and joint torque;

[0009] S4, substituting the data obtained in step S3 into the identification model, directly removing specific columns in the observation matrix to make it full rank according to the minimum parameter set, calculating a set of initial solutions by the least squares method, using a certain range of the initial solutions as the search range of the particle swarm optimization algorithm for adaptive mutation, and running the algorithm to identify the dynamic parameters of the robot;

[0010] S5. Drive the robot to run the verification trajectory, process the collected data to obtain the observation matrix of the verification trajectory, and after full-rank processing, bring the dynamic parameters identified with the excitation trajectory into the linear model to obtain the calculated torque. Compare the calculated torque with the collected torque to verify the accuracy of the identified parameters.

[0011] Furthermore, in the step S1, based on the robot DH parameters including connecting rod length, connecting rod torsion angle, offset and joint angle, the conversion matrix between the coordinate systems is determined, and the dynamic model of the robot is established through the Newton-Euler dynamic equation. After linearizing the dynamic model, a dynamic linear model containing a dynamic parameter set is obtained, and the minimum dynamic equation is: In this formula, τ is the moment of each joint, is the coefficient regression matrix, and P is the minimum parameter set of the robot.

[0012] Furthermore, the mixed excitation trajectory expression in step S2 is: In the above formula, q(t) represents the joint position, q0 is the initial position vector, and the initial position of the robot when running the excitation trajectory can be changed by changing q0.

[0013] Furthermore, in step S4, the least squares and adaptive mutation particle swarm algorithm are used to identify parameters. First, a set of initial solutions are obtained by using the least squares estimation in MATLAB: ls =(Y T Y) -1 Y T τ, according to the initial solution P ls Determine the search range of the particle swarm algorithm so that particles can refine the solution near the initial solution.

[0014] The present invention provides an industrial robot dynamic parameter identification method based on an adaptive mutation particle swarm algorithm. Compared with the prior art, it has the following beneficial effects:

[0015] 1. The method of the minimum parameter set adopted by the present invention is not limited to the number and type of joints of the robot, and can quickly make the observation matrix full rank to obtain the coefficient regression matrix, thereby improving the efficiency of trajectory optimization and identification, and can identify all the parameters of the robot at one time, quickly and accurately, and can also be applied to the identification of different friction models and loads;

[0016] 2. The present invention introduces an initial solution to reduce the search range of the particle swarm algorithm, improves the recognition efficiency, introduces an adaptive mutation operation, helps the particle swarm algorithm to jump out of the local optimal solution, and increases the diversity of the algorithm;

[0017] 3. The present invention enables the robot to start moving from different starting positions, which is convenient for trajectory selection and iterative verification experiments. At the same time, the recognition accuracy is high, the predicted torque follow-up is good, and the torque error root mean square is small. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0019] Figure 1 Iterative diagram of the present invention running the adaptive mutation particle swarm algorithm to identify the dynamic parameters;

[0020] Figure 2 After the industrial robot of the present invention runs the excitation trajectory, the identified dynamic parameters are fed back into the dynamic model, and the calculated torque obtained is compared with the torque collected by the excitation trajectory;

[0021] Figure 3 After the industrial robot of the present invention runs the verification trajectory, the identified dynamic parameters are brought into the dynamic model, and the calculated torque obtained is compared with the torque collected by the verification trajectory; DETAILED DESCRIPTION

[0022] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention are clearly and completely described. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0023] Example

[0024] In order to solve the technical problems in the background technology, the following industrial robot dynamic parameter identification method based on adaptive mutation particle swarm algorithm is given:

[0025] Combination Figure 1-3 As shown, the method for identifying the dynamic parameters of an industrial robot using an adaptive mutation particle swarm algorithm provided by the present invention comprises the following steps:

[0026] S1. Establish a dynamic model based on the robot's DH parameters and Newton-Euler dynamic equations, linearize the dynamic model, and directly determine the minimum parameter set based on the DH parameters and joint types;

[0027] In step S1, based on the robot DH parameters including connecting rod length, connecting rod torsion angle, offset and joint angle, the conversion matrix between coordinate systems is determined, and the dynamic model of the robot is established through the Newton-Euler dynamic equation. After linearizing the dynamic model, a dynamic linear model containing a set of dynamic parameters is obtained. According to the robot structure and DH parameters, a set of minimum inertia parameters can be directly determined. After removing the corresponding redundant columns in the observation matrix, a coefficient regression matrix with full column rank is obtained, and its minimum dynamic equation is: In this formula, τ is the moment of each joint, is the coefficient regression matrix, P is the minimum parameter set of the robot, τ and It can be obtained by experimental measurement. The solution of P requires an identification algorithm. The present invention proposes a least squares and adaptive mutation particle swarm algorithm to identify the parameters. It should be noted that the friction parameter is only added after the minimum parameter set. For example, the parameter set after adding the friction model is: Among them, P n is the original minimum parameter set of the robot, P f is the friction parameter set after adding the friction model. Correspondingly, the torque corresponding to the linear model is The corresponding coefficient regression matrix is

[0028] S2. Optimize two hybrid trajectories of Fourier series and polynomials in MATLAB, one as the excitation trajectory and the other as the verification trajectory;

[0029] The expression of the mixed excitation trajectory in step S2 is: In the above formula, q(t) represents the joint position, q0 is the initial position vector, and the initial position of the robot when running the excitation trajectory can be changed by changing q0;

[0030] is the traditional finite-term Fourier series, which ensures the periodicity of the excitation trajectory, where N is the Fourier order, A k and B k is the coefficient to be optimized, ω f is the fundamental frequency of the Fourier series, C0+C1t+C2t 2 +C3t 3 +C4t 4 +C5t 5 is a fifth-order polynomial, whose coefficients are specifically the following formula, where t e is the end time of the trajectory cycle:

[0031]

[0032] The optimization of the trajectory constrains the excitation trajectory according to the joint design restrictions of the robot and the speed and angular velocity at the start and stop times to be 0. The optimization criterion is to minimize the condition number cond(Y) of the regression matrix, which can reduce the morbidity of the matrix and increase the anti-noise ability of the excitation trajectory. The constraints include linear equality and nonlinear inequality constraints. In MATLAB, the optimization toolbox patternsearch or fmincon can be used for constrained optimization. By changing q0, at least two trajectories can be obtained, one for the excitation trajectory and the other for the verification trajectory.

[0033] S3, based on step S2, drive the robot to run step S2 to optimize the excitation trajectory, collect data and filter and calculate to obtain the observation matrix and joint torque;

[0034] Since the collected acceleration has large noise, the acceleration is obtained by differentiating the filtered velocity. After data processing, the coefficient regression matrix and torque of the robot can be obtained.

[0035] S4, substituting the data obtained in step S3 into the identification model, directly removing specific columns in the observation matrix to make it full rank according to the minimum parameter set, calculating a set of initial solutions by the least squares method, using a certain range of the initial solutions as the search range of the particle swarm optimization algorithm for adaptive mutation, and running the algorithm to identify the dynamic parameters of the robot;

[0036] In step S4, the least squares and adaptive mutation particle swarm algorithm are used to identify parameters. First, a set of initial solutions are obtained using the least squares estimation in MATLAB: ls =(Y T Y) -1 Y T τ, according to the initial solution P ls Determine the search range of the particle swarm algorithm so that the particles can refine the solution near the initial solution, which can reduce the algorithm calculation time and improve the recognition efficiency;

[0037] The present invention proposes a particle swarm parameter identification algorithm that introduces adaptive variation to particles. For the i-th particle, its speed and position are updated according to the rules of the standard particle swarm algorithm.

[0038] v i (t+1)=ω·v i (t)+c1·rand()·(pbest i -x i (t))+c2·rand()·(gbest-x i (t))

[0039] x i (t+1)=x i (t)+v i(t+1)

[0040] v and x are the particle velocity and position, t is the number of iterations, ω is the inertia weight, c1 and c2 are the individual learning factor and the social learning factor, rand() is a random number, pbest i is the best historical position of each particle, and gbest is the global best historical position. The adaptive mutation operation of the particle position is as follows: decide whether to perform the mutation operation on the particle with a certain probability m. If it is decided to perform the mutation operation, select a dimension n, and mutate the value of the particle position on the dimension according to the selected dimension. If the dimension reaches the preset value a, the value on the dimension is set to 10×rand(), otherwise, the value on the dimension is set to 5×rand(). After iterative optimization, the minimum parameter set P of the robot is finally obtained.

[0041]

[0042] In this formula, n is equal to the preset value a and n is equal to other values ​​to perform different position mutation processing, and rand() is a random number between (0,1). i,n This is the updated particle swarm position after adaptive mutation. It should be noted that the mutation operation can also be applied to particle velocity. The cost function is the sum of squares of the errors between the calculated torque and the true torque. The minimum parameter set P of the robot is finally obtained through iterative optimization.

[0043] S5, drive the robot to run the verification trajectory, process the collected data to obtain the observation matrix of the verification trajectory, and after full rank processing, bring the dynamic parameters identified with the excitation trajectory into the linear model to obtain the calculated torque, compare the calculated torque with the collected torque to verify the accuracy of the identified parameters, and use the verification trajectory optimized in step S2 to verify the experiment. Drive the robot to run the verification trajectory, collect data, process it, and calculate the torque through the dynamic model with the minimum parameter set identified in step S4, and compare it with the experimental torque collected by the verification trajectory to verify the accuracy of the identified parameters, using the root mean square (RMS) of the torque error as the evaluation index.

[0044] In summary, the present invention introduces an initial solution to reduce the search range of the particle swarm algorithm and improve the identification efficiency, introduces an adaptive mutation operation to help the particle swarm algorithm jump out of the local optimal solution and increase the diversity of the algorithm; at the same time, the method of the minimum parameter set is not limited to the number and type of joints of the robot, and the observation matrix can be quickly made full rank to obtain the coefficient regression matrix, thereby improving the efficiency of trajectory optimization and identification, and all parameters of the robot can be identified quickly and accurately at one time, and can also be applied to the identification of different friction models and loads.

[0045] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0046] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. Industrial robot dynamic parameter identification method based on adaptive mutation particle swarm algorithm, Features: The steps include: S1. Establish a dynamic model based on the robot's DH parameters and Newton-Euler dynamic equations, linearize the dynamic model, and directly determine the minimum parameter set based on the DH parameters and joint types; S2. Optimize two hybrid trajectories of Fourier series and polynomials in MATLAB, one as the excitation trajectory and the other as the verification trajectory; S3, based on step S2, drive the robot to run step S2 to optimize the excitation trajectory, collect data and filter and calculate to obtain the observation matrix and joint torque; S4, substituting the data obtained in step S3 into the identification model, directly removing specific columns in the observation matrix to make it full rank according to the minimum parameter set, calculating a set of initial solutions by the least squares method, using a certain range of the initial solutions as the search range of the particle swarm optimization algorithm for adaptive mutation, and running the algorithm to identify the dynamic parameters of the robot; S5. Drive the robot to run the verification trajectory, process the collected data to obtain the observation matrix of the verification trajectory, and after full-rank processing, bring the dynamic parameters identified with the excitation trajectory into the linear model to obtain the calculated torque. Compare the calculated torque with the collected torque to verify the accuracy of the identified parameters.

2. The method for identifying the dynamic parameters of an industrial robot using the adaptive mutation particle swarm algorithm according to claim 1, Features: In the step S1, based on the robot DH parameters including connecting rod length, connecting rod torsion angle, offset and joint angle, the conversion matrix between the coordinate systems is determined, and the dynamic model of the robot is established through the Newton-Euler dynamic equation. After linearizing the dynamic model, a dynamic linear model containing a dynamic parameter set is obtained, and the minimum dynamic equation is: In this formula, τ is the moment of each joint, is the coefficient regression matrix, and P is the minimum parameter set of the robot.

3. The method for identifying dynamic parameters of industrial robots using the adaptive mutation particle swarm algorithm according to claim 1, Features: The mixed excitation trajectory expression in step S2 is: In the above formula, q(t) represents the joint position, q 0 is the initial position vector, which can be changed by changing q 0 To change the initial position of the robot when it runs the excitation trajectory.

4. The method for identifying dynamic parameters of industrial robots using the adaptive mutation particle swarm algorithm according to claim 1, Features: In the step S4, the least squares and adaptive mutation particle swarm algorithm are used to identify parameters. First, a set of initial solutions is obtained by least squares estimation in MATLAB: ls =(Y T Y) -1 Y T τ, according to the initial solution P ls Determine the search range of the particle swarm algorithm so that particles can refine the solution near the initial solution.

Images