Friction identification method based on moss growth algorithm

Abstract

The invention belongs to the technical field of robots, and particularly relates to a friction identification method based on a moss growth algorithm. The method comprises the steps of 1, designing a movement track; 2, executing the motion trail in the step 1 in a position control mode, collecting the speed and moment data of the robot, and extracting the friction moment of the robot; step 3, filtering the collected friction torque; 4, establishing a friction force and Coriolis force coupling model; 5, identifying parameters of the friction force and Coriolis force coupling model by using a moss growth algorithm; and 6, removing the Coriolis force model to obtain a final friction model. According to the method, the problem that a traditional genetic algorithm is prone to falling into a local optimal solution in the parameter identification process is solved, and the identification efficiency and precision are improved.

Description

Technical Field

[0001] The invention belongs to the field of robot technology, in particular to a friction identification method based on a moss growth algorithm. Background Art

[0002] In the fourth industrial revolution, where artificial intelligence and machine automation play a core role, robots occupy an important strategic position. The application of robots in the manufacturing industry, especially in the human-machine collaboration scenario, has posed further challenges to the collection or estimation of the forces of interaction between robots and humans. Sometimes, due to the difficulty and cost of sensor integration, sensorless external force estimation methods are used in human-machine interaction to estimate the interaction forces between robots and the physical world. However, this external force estimation method is often affected by the uncertainty of robot friction, so accurate modeling of friction is necessary. Among them, the selection and establishment of mathematical friction models and the identification of mathematical friction model parameters are the key technologies for robot friction modeling. However, the current friction modeling has the following technical problems: the traditional strbeck friction model is discontinuous at zero crossing, and the resulting discontinuous voltage servo mechanism is difficult to support, which will affect the actual realization of the control signal; in the process of extracting friction force, the coupling of the Coriolis force of a single joint uniform reciprocating motion at high speed is difficult to separate; the linear least squares method is difficult to identify nonlinear friction models, and the traditional genetic algorithm is prone to fall into local optimal solutions. Summary of the invention

[0003] To solve the above problems, the present invention provides a friction identification method based on moss growth algorithm, which can improve the continuity of the friction model, avoid the problem of easily falling into the local optimal solution in the traditional genetic algorithm parameter identification process, and improve the identification efficiency and accuracy.

[0004] The technical solution of the present invention is described as follows in conjunction with the accompanying drawings:

[0005] A friction identification method based on moss growth algorithm comprises the following steps:

[0006] Step 1: Design a motion trajectory;

[0007] Step 2: Execute the motion trajectory of step 1 in position control mode, collect the speed and torque data of the robot, and extract the friction torque of the robot;

[0008] Step 3: filtering the collected friction torque;

[0009] Step 4: Establish a coupling model of friction and Coriolis force;

[0010] Step 5: Using moss growth algorithm to identify the parameters of the friction and Coriolis force coupling model;

[0011] Step 6: Eliminate the Coriolis force model to obtain the final friction model.

[0012] Furthermore, the specific method of step one is as follows:

[0013] 11) For any joint i of the robot, make any joint do single-joint uniform reciprocating motion; that is, lock the other joints and make joint i rotate at a uniform speed;

[0014] 12) The speed range for uniform motion includes two speed ranges, low speed and high speed. The low speed range should have more speed sampling points than the high speed range. The joint motion range should be based on fully stimulating the robot's motion characteristics and not exceeding the robot's joint motion limit, and move in both clockwise and counterclockwise directions.

[0015] Furthermore, the specific method of step 2 is as follows:

[0016] 21) The speed and torque data of the robot are collected from the motor encoder position differential and the motor torque output current sampling, and converted into the motor output torque by the following formula:

[0017] τ m =I m K t N m (1)

[0018] In the formula, τ m is the motor equivalent output torque; I m is the motor sampling current; K t is the torque constant of the servo motor; N m is the single joint transmission ratio;

[0019] 22) Calculate the friction torque of the robot and extract it using the following formula:

[0020]

[0021] In the formula, τ m1 and τ m2 are the joint torques collected from the clockwise and counterclockwise motions of a single joint of the robot; τ f is the extracted friction torque; is the joint angular velocity of the robot.

[0022] Furthermore, the specific method of step three is as follows:

[0023] The collected friction torque is filtered using Butterworth filtering.

[0024] Furthermore, the specific method of step 4 is as follows:

[0025] The coupling model of friction and Coriolis force is established as follows:

[0026]

[0027] In the formula, It is the stribeck phenomenon; It is the Coulomb phenomenon. It is a viscosity phenomenon; is the Coriolis force coupling term; where f s1 、f s2 、f s3 、f c1 、f c2 、f v 、f c3 is a diagonal matrix with positive parameters.

[0028] Furthermore, the specific method of step five is as follows:

[0029] 51) The evolutionary direction of a population is determined by a mechanism called wind direction determination, which uses a method of dividing the population;

[0030] 52) Two new reproduction strategies, spore diffusion reproduction and double reproduction, are proposed for exploration and development;

[0031] 53) The hidden mechanism is used to modify individual solutions to avoid the algorithm falling into local optimality.

[0032] Furthermore, the specific method of step 51) is as follows:

[0033] The wind direction determination mechanism expression is given by the following algorithm:

[0034]

[0035] Where M best is the best moss individual in population X, and the j-th dimension value of all individuals in the population is multiplied by M best The j-th dimension value is compared to divide the population into DX j1 and DX j2 Two parts:

[0036] And compare the sets with the largest number, expressed as follows:

[0037]

[0038] In the formula, count is the number of moss individuals counted;

[0039] After multiple splits, the resulting set is:

[0040]

[0041] Where, d n For the number of divisions to be made, take d n = dim / 4 and d n Not less than 1; p j is the jth random number, ranging from (1,2,......,dim), and conforms to the equation:

[0042]

[0043] The wind direction always tends to blow from divX to M best Region, define wind direction as:

[0044]

[0045] Where dirX is defined as the individual pointer M in divX. best The set of vectors is expressed as follows:

[0046] dirX={M best -M i |M i ∈divX}, (9).

[0047] Furthermore, the specific method of step 52) is as follows:

[0048] The spore diffusion search for spore diffusion reproduction is given by the following expression:

[0049]

[0050] In the formula, is the new moss individual obtained by the diffusion of the ith moss; r1 is a random number in the range of (0,1); d1 is a constant parameter, set to 0.2;

[0051] Spores travel a long distance in steady wind conditions and a short distance in turbulent wind conditions. The step lengths are given by:

[0052] step1=w·(r2-0.5)·E, (11)

[0053]

[0054] Where r2 and r3 are random numbers between (0, 1); E is the wind intensity; β is the ratio of the number of individuals in divX to the total number of individuals in X, which are given by the following formulas:

[0055]

[0056] Where FEs is the current number of iterations; MaxFEs is the maximum number of iterations;

[0057] The bidirectional propagation of double reproduction is given by the following expression:

[0058]

[0059] In the formula, is the i-th new individual; for The jth particle in ; j is a random integer not exceeding the maximum dimension of the individual; r4 is a random number in the range (0,1); d2 is a constant, which is set to 0.5; act and step3 are given by the following expressions:

[0060]

[0061] step3=0.1·(r6-0.5)·E, (17)

[0062] In the formula, r5 is a random number between (0,1) and M best Same dimensions.

[0063] Further, the specific method of 53) is as follows:

[0064] Every ten mutations of a moss individual Mi triggers the cryptobiotic search mechanism, searching for the best individual among the historically mutated moss individuals. Finally, the amplitude of the best individual is given to the moss individual Mi, and subsequent operations are continued.

[0065] Among them, create moss individuals, the dimension is equal to the parameter matrix f s1 、f s2 、f s3 、f c1 、f c2 、f v 、f c3 Dimension, the dimension is 7, and the fitness function is defined as:

[0066]

[0067] Where S is the fitness value of the moss individual; L fi is the actual value of the friction torque and Coriolis force coupling of the i-th group; MGO stbi is the estimated value of the model of the i-th group; n is the number of speed sampling groups;

[0068] Furthermore, the specific method of step six is ​​as follows:

[0069] 62) Eliminating the Coriolis force model is expressed as setting q=0 in formula (3), thus obtaining a pure friction model.

[0070] The beneficial effects of the present invention are:

[0071] 1) In terms of robot friction modeling, the present invention conducts targeted modeling and identification based on the different friction characteristics of the robot in different speed ranges, solving the problem that the stribeck friction model is discontinuous at the zero crossing and the coupling of the Coriolis force and the inertial force at high speeds causes errors in friction collection.

[0072] 2) In terms of robot friction model identification, the present invention targets the nonlinear characteristics of the robot friction model. It is impossible to obtain accurate model parameters using linear identification methods. Traditional identification methods such as genetic algorithms have the problems of slow identification speed and easy to fall into local optimal solutions. The moss growth optimization algorithm is applied to friction identification, which improves the identification efficiency and avoids the problem of easy falling into local optimal solutions during the identification process. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0074] Figure 1 It is the overall flow chart of the present invention;

[0075] Figure 2 Flowchart of cryptobiotic mechanism for moss growth algorithm;

[0076] Figure 3 Flowchart of the moss growth algorithm;

[0077] Figure 4 This is the effect diagram of robot friction model identification. DETAILED DESCRIPTION

[0078] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.

[0079] Embodiment 1

[0080] See also Figure 1 This embodiment provides a friction identification method based on a moss growth algorithm, comprising the following steps:

[0081] Step 1: Design a motion trajectory, as follows:

[0082] 11) For any joint i of the robot, make any joint do single-joint uniform reciprocating motion; that is, lock the other joints and make joint i rotate at a uniform speed;

[0083] 12) The speed range of uniform motion should be as sufficient as possible, including both low-speed and high-speed ranges. More speed sampling points should be set in the low-speed range. The joint motion range should be based on fully stimulating the robot's motion characteristics and not exceed the robot's joint motion limit, and move in both clockwise and counterclockwise directions.

[0084] Step 2: Execute the motion trajectory of step 1 in position control mode, collect the speed and torque data of the robot, and extract the friction torque of the robot, as follows:

[0085] 21) The speed and torque data of the robot are collected from the motor encoder position differential and the motor torque output current sampling, and converted into the motor output torque by the following formula:

[0086] τ m =I m K t N m (1)

[0087] In the formula, τ m is the motor equivalent output torque; I m is the motor sampling current; K t is the torque constant of the servo motor; N m is the single joint transmission ratio;

[0088] 22) Calculate the friction torque of the robot and extract it using the following formula:

[0089]

[0090] In the formula, τ m1 and τ m2 are the joint torques collected from the clockwise and counterclockwise motions of a single joint of the robot; τ f is the extracted friction torque; is the joint angular velocity of the robot.

[0091] Step 3: Filter the collected friction torque, as follows:

[0092] The collected friction torque is filtered using Butterworth filtering.

[0093] Step 4: Establish a coupling model of friction and Coriolis force, as follows:

[0094] The coupling model of friction and Coriolis force is established as follows:

[0095]

[0096] In the formula, It is the stribeck phenomenon; It is the Coulomb phenomenon. It is a viscosity phenomenon; is the Coriolis force coupling term; where f s1 、f s2 、f s3 、f c1 、f c2 、f v 、f c3 is a diagonal matrix with positive parameters.

[0097] Step 5: Read Figure 3 , the parameters of the friction and Coriolis force coupling model are identified using the moss growth algorithm; the moss growth algorithm is a metaheuristic algorithm that first determines the evolutionary direction of the population through a mechanism called wind direction determination, which uses a method of dividing the population. At the same time, drawing on the experience of sexual and asexual reproduction of mosses, two new reproduction strategies, namely spore diffusion reproduction and double reproduction, are proposed for exploration and development. Finally, the cryptobacterial mechanism changes the traditional metaheuristic algorithm's method of directly modifying individual solutions, avoiding the algorithm from falling into local optimality, as follows:

[0098] 51) The evolutionary direction of a population is determined by a mechanism called wind direction determination, which uses a method of dividing the population;

[0099] The wind direction determination mechanism expression is given by the following algorithm:

[0100]

[0101] Where M best is the best moss individual in population X, and the j-th dimension value of all individuals in the population is multiplied by M best The jth dimension value is compared to divide the population into DX j1 and DX j2 Two parts:

[0102] And compare the sets with the largest number, expressed as follows:

[0103]

[0104] In the formula, count is the number of moss individuals counted;

[0105] After multiple splits, the resulting set is:

[0106]

[0107] Where, d n For the number of divisions to be made, take d n = dim / 4 and d n Not less than 1; p j is the jth random number, ranging from (1,2,......,dim), and conforms to the equation:

[0108]

[0109] The wind direction always tends to blow from divX to M best Region, define wind direction as:

[0110]

[0111] Where dirX is defined as the individual pointer M in divX. best The set of vectors is expressed as follows:

[0112] dirX={M best -M i |M i ∈divX}, (9).

[0113] 52) Two new reproduction strategies, spore diffusion reproduction and double reproduction, are proposed for exploration and development;

[0114] The spore diffusion search for spore diffusion reproduction is given by the following expression:

[0115]

[0116] In the formula, is the new moss individual obtained by the diffusion of the ith moss; r1 is a random number in the range of (0,1); d1 is a constant parameter, set to 0.2;

[0117] Spores travel a long distance in steady wind conditions and a short distance in turbulent wind conditions. The step lengths are given by:

[0118] step1=w·(r2-0.5)·E, (11)

[0119]

[0120] Where r2 and r3 are random numbers between (0, 1); E is the wind intensity; β is the ratio of the number of individuals in divX to the total number of individuals in X, which are given by the following formulas:

[0121]

[0122] Where FEs is the current number of iterations; MaxFEs is the maximum number of iterations;

[0123] The bidirectional propagation of double reproduction is given by the following expression:

[0124]

[0125] In the formula, is the i-th new individual; for The jth particle in ; j is a random integer not exceeding the maximum dimension of the individual; r4 is a random number in the range (0,1); d2 is a constant, which is set to 0.5; act and step3 are given by the following expressions:

[0126]

[0127] step3=0.1·(r6-0.5)·E, (17)

[0128] In the formula, r5 is a random number between (0,1) and M best Same dimensions.

[0129] 53) The hidden mechanism is used to modify the individual solution to avoid the algorithm falling into the local optimum;

[0130] Every ten mutations of a moss individual Mi triggers the cryptobiotic search mechanism, searching for the best individual among the historically mutated moss individuals. Finally, the amplitude of the best individual is given to the moss individual Mi, and subsequent operations are continued.

[0131] Figure 2 It is a flowchart of the specific algorithm, where record represents the number of iterations of the record, Mi is the i-th individual of the moss population, rMi_record represents the Mi mutant individual of the record generation, rec_num is the maximum value of record, T is the maximum number of iterations, and rMi_best is the best generation of individuals in this cryptobiotic mechanism search.

[0132] Among them, create moss individuals, the dimension is equal to the parameter matrix f s1 、f s2 、f s3 、f c1 、f c2 、f v 、f c3 Dimension, the dimension is 7, and the fitness function is defined as:

[0133]

[0134] Where S is the fitness value of the moss individual; L fi is the actual value of the friction torque and Coriolis force coupling of the i-th group; MGO stbi is the estimated value of the model of the i-th group; n is the number of speed sampling groups;

[0135] Step 6: Eliminate the Coriolis force model to obtain the final friction model, as follows:

[0136] Eliminating the Coriolis force modeling is expressed as setting q=0 in formula (3), thus obtaining a pure friction model.

[0137] Embodiment 2

[0138] This embodiment experimentally verifies the robot friction model parameter identification method based on the improved moss growth algorithm proposed in the first embodiment.

[0139] The robot collects position, speed, and torque information at a speed of -2rad / s to 2rad / s for a single joint. The parameter identification effect is as follows: Figure 4 shown.

[0140] The above experimental results show that the friction identification method based on the moss growth algorithm proposed in the present invention can effectively identify the friction model of the robot, and the model has a good fitting effect.

[0141] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A friction identification method based on moss growth algorithm, characterized in that: The following steps are involved: Step 1: Design a motion trajectory; Step 2: Execute the motion trajectory of step 1 in position control mode, collect the speed and torque data of the robot, and extract the friction torque of the robot; Step 3: filtering the collected friction torque; Step 4: Establish a coupling model of friction and Coriolis force; Step 5: Using moss growth algorithm to identify the parameters of the friction and Coriolis force coupling model; Step 6: Eliminate the Coriolis force model to obtain the final friction model.

2. The friction identification method based on moss growth algorithm according to claim 1, characterized in that: The specific method of step one is as follows: 11) For any joint i of the robot, make any joint do single-joint uniform reciprocating motion; that is, lock the other joints and make joint i rotate at a uniform speed; 12) The speed range for uniform motion includes two speed ranges, low speed and high speed. The low speed range should have more speed sampling points than the high speed range. The joint motion range should be based on fully stimulating the robot's motion characteristics and not exceeding the robot's joint motion limit, and move in both clockwise and counterclockwise directions.

3. The friction identification method based on moss growth algorithm according to claim 1 is characterized in that: The specific method of step 2 is as follows: 21) The speed and torque data of the robot are collected from the motor encoder position differential and the motor torque output current sampling, and converted into the motor output torque by the following formula: t m =I m K t N m (1) In the formula, τ m is the motor equivalent output torque; I m is the motor sampling current; K t is the torque constant of the servo motor; N m It is the single joint transmission ratio; 22) Calculate the friction torque of the robot and extract it using the following formula: In the formula, τ m1 and τ m2 are the joint torques collected from the clockwise and counterclockwise motions of a single joint of the robot; τ f is the extracted friction torque; is the joint angular velocity of the robot.

4. The friction identification method based on moss growth algorithm according to claim 1, characterized in that: The specific method of step three is as follows: The collected friction torque is filtered using Butterworth filtering.

5. The friction identification method based on moss growth algorithm according to claim 1, characterized in that: The specific method of step 4 is as follows: The coupling model of friction and Coriolis force is established as follows: In the formula, It is the stribeck phenomenon; It is the Coulomb phenomenon. It is a viscosity phenomenon; is the Coriolis force coupling term; where f s1 、f s2 、f s3 、f c1 、f c2 、f v 、f c3 is a diagonal matrix with positive parameters.

6. The friction identification method based on moss growth algorithm according to claim 1, characterized in that: The specific method of step five is as follows: 51) The evolutionary direction of a population is determined by a mechanism called wind direction determination, which uses a method of dividing the population; 52) Two new reproduction strategies, spore diffusion reproduction and double reproduction, are proposed for exploration and development; 53) The hidden mechanism is used to modify individual solutions to avoid the algorithm falling into local optimality.

7. The friction identification method based on moss growth algorithm according to claim 6, characterized in that: The specific method of step 51) is as follows: The wind direction determination mechanism expression is given by the following algorithm: Where M best is the best moss individual in population X, and the j-th dimension value of all individuals in the population is multiplied by M best The jth dimension value is compared to divide the population into DX j1 and DX j2 Two parts: And compare the sets with the largest number, expressed as follows: In the formula, count is the number of moss individuals counted; After multiple splits, the resulting set is: Where, d n For the number of divisions to be made, take d n = dim / 4 and d n Not less than 1; p j is the jth random number, ranging from (1,2,......,dim), and conforms to the equation: The wind direction always tends to blow from divX to M best Region, define wind direction as: Where dirX is defined as the individual pointer M in divX. best The set of vectors is expressed as follows: dirX={M best -M i |M i ∈divX}, (9).

8. The friction identification method based on moss growth algorithm according to claim 6, characterized in that: The specific method of step 52) is as follows: The spore diffusion search for spore diffusion reproduction is given by the following expression: In the formula, is the new moss individual obtained by the diffusion of the ith moss; r1 is a random number in the range of (0,1); d1 is a constant parameter, set to 0.2; Spores travel a long distance in steady wind conditions and a short distance in turbulent wind conditions. The step lengths are given by: step1=w·(r2-0.5)·E, (11) Where r2 and r3 are random numbers between (0, 1); E is the wind intensity; β is the ratio of the number of individuals in divX to the total number of individuals in X, which are given by the following formulas: Where FEs is the current number of iterations; MaxFEs is the maximum number of iterations; The bidirectional propagation of double reproduction is given by the following expression: In the formula, is the i-th new individual; for The jth particle in ; j is a random integer not exceeding the maximum dimension of the individual; r4 is a random number in the range (0,1); d2 is a constant, which is set to 0.5; act and step3 are given by the following expressions: step3=0.1·(r6-0.5)·E, (17) In the formula, r5 is a random number between (0,1) and M best Same dimensions.

9. The friction identification method based on moss growth algorithm according to claim 6, characterized in that: The specific method of the above 53) is as follows: Every ten mutations of a moss individual Mi triggers the cryptobiotic search mechanism, searching for the best individual among the historically mutated moss individuals, and finally giving the obtained best individual amplitude to the moss individual Mi, and continuing to perform subsequent operations; Among them, create moss individuals, the dimension is equal to the parameter matrix f s1 、f s2 、f s3 、f c1 、f c2 、f v 、f c3 Dimension, the dimension is 7, and the fitness function is defined as: Where S is the fitness value of the moss individual; L fi is the actual value of the friction torque and Coriolis force coupling of the i-th group; MGO stbi is the estimated value of the model of the i-th group; n is the number of speed sampling groups.

10. The friction identification method based on moss growth algorithm according to claim 5, characterized in that: The specific method of step six is ​​as follows: Eliminating the Coriolis force modeling is expressed as setting q=0 in formula (3), thus obtaining a pure friction model.

Images