A robot gripper structure parameter rapid simulation optimization method

Abstract

The application belongs to the field of rapid design and manufacture of production equipment, and particularly relates to a robot gripper structure parameter rapid simulation optimization method, which comprises the following steps: constructing a robot gripper structure parameterization model; simulating the stress and deformation of the robot gripper structure parameterization model under different sizes and different load conditions by using statics finite element simulation software to obtain stress and deformation simulation data; constructing a stress and deformation prediction model; optimizing the size parameters that meet the stress and deformation constraint conditions under the actual load conditions by establishing a particle swarm optimization algorithm, obtaining the optimal value of the size parameters, and substituting the optimal value into the stress and deformation prediction model to obtain the size parameters under the actual conditions and the stress and deformation under the load, so as to realize the optimization of the robot gripper structure parameters. The application utilizes the Kriging surrogate model to predict the stress and deformation, reduces the calling of the finite element model, and improves the optimization efficiency.

Description

Technical Field

[0001] This invention belongs to the field of rapid design and manufacturing of production equipment, specifically a rapid simulation and optimization method for the structural parameters of a robot gripper. Background Technology

[0002] With the rapid development of customization in manufacturing, the industry is gradually shifting from large-scale assembly line production to small-batch, customized production. Therefore, automation is the future direction of smart workshops, which places higher demands on the design and manufacturing of production lines, requiring greater flexibility and adaptability. Gantry robots are a widely used type of production equipment, extensively used in areas such as material handling and welding.

[0003] In the optimization design of mechanical structures, due to the large number of combinations of loads and dimensions of components, the traditional method of performing finite element analysis on stress, deformation, etc. to determine whether they meet the operating conditions would waste a lot of time and reduce design efficiency. Summary of the Invention

[0004] The purpose of this invention is to provide a rapid simulation and optimization method for robot gripper structures, so as to overcome the problem that the stress, deformation, etc. of the robot gripper structure parameters must be analyzed by finite element analysis every time to see if they meet the operating conditions.

[0005] The technical solution adopted by this invention to achieve the above objectives is: a method for rapid simulation and optimization of robot gripper structural parameters, comprising the following steps:

[0006] Step 1: Construct a parametric model of the robot gripper structure to provide a model for stress and deformation in static finite element simulation;

[0007] Step 2: Use static finite element simulation software to simulate the stress and deformation of the parameterized model of the robot gripper structure under different dimensions and load conditions, and obtain stress and deformation simulation data.

[0008] Step 3: Based on the obtained stress and deformation simulation data, construct a stress and deformation prediction model for the robot gripper containing the structural size parameters of the robot gripper and under different load conditions using a Kriging surrogate model;

[0009] Step 4: By establishing a particle swarm optimization algorithm, the dimensional parameters that meet the stress and deformation constraints under actual load conditions are optimized. The optimal values ​​of the dimensional parameters are then substituted into the stress and deformation prediction model to obtain the dimensional parameters under actual working conditions, as well as the stress and deformation under load, thereby optimizing the structural parameters of the robot gripper.

[0010] Step 1 includes the following steps:

[0011] Step 1.1: Construct a 3D model of the robot gripper structure using 3D modeling software;

[0012] Step 1.2: Input the wall thickness DH and wall spacing DV of the robot gripper structure into the 3D model of the robot gripper structure to obtain the parametric model of the robot gripper structure.

[0013] Step 2 includes the following steps:

[0014] Step 2.1: In the static finite element simulation software, perform hexahedral mesh generation on the parametric model of the robot gripper structure;

[0015] Step 2.2: In the static finite element simulation software, loads F under different working conditions are applied to different dimensional parameters DH and DV to perform static simulation and obtain stress and deformation simulation data.

[0016] Step 2.2 specifically includes:

[0017] (1) Based on the parameterized model of the robot gripper structure after the hexahedral mesh division in step 2.1, a fixed constraint is added to the robot gripper structure to indicate that the upper end of the robot gripper is tightly connected to the gripper base;

[0018] (2) The lower end of the robot gripper contacts the target object, and the gripping force is provided by the cylinder. The weight of the target object and the gripping force of the cylinder are decomposed into two contact surfaces using the trigonometric function decomposition method, that is:

[0019]

[0020]

[0021] Among them, F 夹 denoted as the clamping force of the cylinder, m as the mass of the target being clamped, g as the acceleration due to gravity, R as the angle between the grippers, and n as the number of gripping arms.

[0022] (3) Loads under different working conditions are applied sequentially to DH and DV with different dimensional parameters. A and F B Static simulation was performed using ANSYS Workbench software to obtain stress and deformation simulation data.

[0023] Step 3 includes the following steps:

[0024] Step 3.1: Sample the stress and deformation simulation data obtained in Step 2 using Latin hypercube sampling;

[0025] Step 3.2: Construct a dataset using all the sampled stress and deformation simulation data obtained in Step 3.1, and establish a Kriging surrogate calculation model for the robot gripper containing the structural dimensional parameters of the robot gripper and different load conditions, i.e., a stress and deformation prediction model.

[0026] Step 3.1 specifically includes:

[0027] The variable space is divided into n non-overlapping sub-intervals with the same probability. In order to ensure the uniformity of the sample points, a random and independent sample is taken in each sub-interval.

[0028] The sample obtained through Latin hypercube sampling is as follows:

[0029]

[0030] In the formula: r i —Number of sample points; i—Number of levels; j—Dimension; d ij —The number of independent permutations of 1 to r; e ij A random number between —.

[0031] Step 3.2 specifically includes:

[0032] The sampled stress and deformation simulation data are used to construct a dataset. The sample data after Latin hypercube sampling is substituted into the Kriging model that has been established in MATLAB to generate a deformation and stress prediction model.

[0033] Using different dimensional parameters and different loads as inputs to the deformation and stress prediction model, and different stresses and deformations as outputs, the resulting deformation and stress prediction model is used as a Kriging proxy calculation model.

[0034] After the Kriging surrogate model is established, the deformation and stress of the gripper component under different dimensional parameters DH, DV and different loads F are predicted using the Kriging surrogate model.

[0035] Step 4 includes the following steps:

[0036] Step 4.1: Establish a particle swarm optimization algorithm. Determine the load based on the actual working conditions and use it as the input parameter for the particle swarm optimization algorithm; use different size parameters as particles.

[0037] Step 4.2: Particle swarm initialization, i.e.: set the population size N; set the number of iterations to M, perform particle swarm optimization to obtain multiple individual optimal values ​​and a population optimal value;

[0038] Step 4.3: Substitute the particles corresponding to the optimal values ​​obtained in Step 4.2 into the Kriging proxy calculation model to obtain the dimensional parameters of the actual working conditions and the stress and deformation under the load;

[0039] Step 4.4: Determine whether the actual working condition dimensional parameters and stress and deformation under load obtained in Step 4.3 meet the set constraints. If they do, optimize the robot gripper structure parameters. If they do not, repeat Steps 1 to 4.

[0040] The particle swarm optimization algorithm is established as follows:

[0041] The mass of the robot gripper structure is used as the fitness function of the particle swarm optimization algorithm;

[0042] The fitness function of the particle swarm optimization algorithm is:

[0043] When the mass M of the beam reaches its minimum, that is:

[0044] M min =M(D1, D2, DH, DV, F1)

[0045] The constraints are set as follows:

[0046]

[0047] Where D1 and D2 are the optimal solutions for a set of design variables; D1 max The maximum value of D1, D2 min The minimum value of D2, Where DH is the allowable stress of the material, DH is the wall thickness of the robot gripper structure, and DV is the wall spacing of the robot gripper structure.

[0048] Take the maximum velocity V of the particle max The corresponding search space is 10% to 20%. The inertial weight ω, the self-learning factor C1, and the social learning factor C2 of the particle swarm algorithm are set to complete the construction of the particle swarm algorithm.

[0049] The present invention has the following beneficial effects and advantages:

[0050] 1. This invention utilizes a Kriging surrogate model to predict stress and deformation, reducing the need to call finite element models and improving optimization efficiency.

[0051] 2. This invention embeds the Kriging surrogate model into the particle swarm optimization algorithm, thereby improving the optimization efficiency of the particle swarm optimization algorithm. Attached Figure Description

[0052] Figure 1 This is the overall flowchart of the rapid simulation optimization method for robot gripper structure parameters of the present invention. Detailed Implementation

[0053] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0054] like Figure 1 The diagram shown is a general flowchart of the rapid simulation optimization method for robot gripper structural parameters according to the present invention. This embodiment discloses a rapid simulation optimization method for robot gripper structural parameters, using a gantry robot as an example, including the following steps:

[0055] Step 1: Construct a parametric model of the gantry robot gripper structure to provide a model for stress and deformation in static finite element simulation;

[0056] Step 2: Use static finite element simulation software to simulate the stress and deformation of the parametric model of the gantry robot gripper structure under different dimensions and load conditions, and obtain stress and deformation simulation data.

[0057] Step 3: Based on the obtained stress and deformation simulation data, construct a stress and deformation prediction model for the gantry robot gripper containing the structural dimensional parameters of the gantry robot gripper and under different load conditions using a Kriging surrogate model.

[0058] Step 4: By establishing a particle swarm optimization algorithm, the dimensional parameters that meet the stress and deformation constraints under actual load conditions are optimized. The optimal values ​​of the dimensional parameters are then substituted into the stress and deformation prediction model to obtain the dimensional parameters under actual working conditions, as well as the stress and deformation under load, thereby optimizing the gripper structure parameters of the gantry robot.

[0059] Specifically, step S1 includes the following steps:

[0060] Step 1.1: Construct a 3D model of the gantry robot gripper structure using 3D modeling software;

[0061] Step 1.2: Input the wall thickness DH and wall spacing DV of the gantry robot gripper structure into the 3D model of the gantry robot gripper structure to obtain the parametric model of the gantry robot gripper structure.

[0062] The parametric design 3D model of the gantry robot gripper obtained in step S1.2 is subjected to stress and deformation static finite element analysis under different dimensional parameters and load conditions. Step S2 includes the following steps:

[0063] Step 2.1: In the static finite element simulation software, perform hexahedral mesh generation on the parametric model of the gantry robot gripper structure;

[0064] Step 2.2: Perform static simulations in static finite element simulation software by applying loads F under different working conditions to different dimensional parameters DH and DV, i.e.:

[0065] (1) Based on the parameterized model of the gantry robot gripper structure after the hexahedral mesh division in step 2.1, a fixed constraint is added to the gantry robot gripper structure to indicate that the upper end of the gantry robot gripper is tightly connected to the gripper base.

[0066] (2) The lower end of the gantry robot's gripper contacts the target object being gripped. The gripping force is provided by a cylinder. The weight of the target object and the gripping force of the cylinder are decomposed using trigonometric functions, distributing the force across two contact surfaces:

[0067]

[0068]

[0069] Among them, F 夹 denoted as the clamping force of the cylinder, m as the mass of the target being clamped, g as the acceleration due to gravity, R as the angle between the grippers, and n as the number of gripping arms.

[0070] (3) Loads under different working conditions are applied sequentially to DH and DV with different dimensional parameters. A and F B Static simulation was performed using ANSYS Workbench software to obtain stress and deformation simulation data.

[0071] Step S3 includes the following steps:

[0072] Step 3.1: Sample the stress and deformation simulation data obtained in Step 2 using Latin hypercube sampling; that is, divide the variable space into n non-overlapping sub-intervals with the same probability. In order to ensure the uniformity of the sample points, perform a random and independent sampling once in each sub-interval.

[0073] The sample obtained through Latin hypercube sampling is as follows:

[0074]

[0075] In the formula: r i —Number of sample points; i—Number of levels; j—Dimension; d ij —The number of independent permutations of 1 to r; e ij A random number between —.

[0076] Step 3.2: Construct a dataset from the sampled stress and deformation simulation data, and substitute the sample set data after Latin hypercube sampling into the Kriging model that has been established in MATLAB to generate a deformation and stress prediction model.

[0077] Using different dimensional parameters and different loads as inputs to the deformation and stress prediction model, and different stresses and deformations as outputs, the resulting deformation and stress prediction model is used as a Kriging proxy calculation model.

[0078] After the Kriging surrogate model is established, the deformation and stress of the gripper component under different dimensional parameters DH, DV and different loads F are predicted using the Kriging surrogate model.

[0079] Step S4 includes the following steps:

[0080] Step 4.1: Establish the particle swarm optimization algorithm;

[0081] The particle swarm optimization algorithm is established as follows:

[0082] The mass of the gripper structure of the gantry robot is used as the fitness function of the particle swarm optimization algorithm;

[0083] The fitness function of the particle swarm optimization algorithm is:

[0084] When the mass M of the beam reaches its minimum, that is:

[0085] M min =M(D1, D2, DH, DV, F1)

[0086] The constraints are set as follows:

[0087]

[0088] Where D1 and D2 are the optimal solutions for a set of design variables; D1 max The maximum value of D1, D2 min The minimum value of D2, DH represents the allowable stress of the material, DH represents the wall thickness of the gantry robot gripper structure, and DV represents the wall spacing of the gantry robot gripper structure.

[0089] The maximum velocity V of the particle is taken. max The corresponding search space is 10% to 20%. In this embodiment, the inertia weight ω of the particle swarm algorithm is set to 0.6, the self-learning factor C1 is set to 2, and the social learning factor C2 is set to 2 to complete the construction of the particle swarm algorithm.

[0090] The load is determined based on the actual working conditions and used as the input parameter for the particle swarm optimization algorithm; different size parameters are used as particles.

[0091] Step 4.2: Initialize the particle swarm optimization, i.e., set the population size N, where N is between 20 and 40; set the number of iterations to 100, perform the particle swarm optimization algorithm to obtain multiple individual optimal values ​​and a population optimal value;

[0092] Step 4.3: Substitute the particles corresponding to the optimal values ​​obtained in Step 4.2 into the Kriging proxy calculation model to obtain the dimensional parameters of the actual working conditions and the stress and deformation under the load;

[0093] Step 4.4: Determine whether the actual working condition dimensional parameters and stress and deformation under load obtained in Step 4.3 meet the set constraints. If they do, optimize the gantry robot gripper structure parameters. If they do not, repeat Steps 1 to 4.

[0094] In summary, this paper proposes an optimization scheme suitable for the structural characteristics of gantry robots, namely, an optimization method combining the Kriging model and the Particle Swarm Optimization (PSO) algorithm. Taking the mass of the gantry robot as the optimization objective, and comprehensively considering constraints such as deformation and load, parametric modeling is employed, and a certain number of data points are selected using the Latin hypercube sampling method to construct the model. The required sample points are obtained through finite element analysis, and a Kriging surrogate model is constructed to replace finite element simulation, reducing the number of finite element simulations and improving optimization efficiency. Finally, the PSO algorithm is used to find the global optimum solution for the Kriging surrogate model. This invention utilizes the Kriging surrogate model to predict stress and deformation, reducing the need to call the finite element model and improving optimization efficiency. Furthermore, embedding the Kriging surrogate model into the PSO algorithm improves the optimization efficiency of the PSO algorithm.

[0095] The above description is merely an embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, extensions, etc., made within the spirit and principles of the present invention are included within the scope of protection of the present invention.

Claims

1. A robot gripper structure parameter rapid simulation optimization method, characterized in that, Includes the following steps: Step 1: Construct a parametric model of the robot gripper structure to provide a model for stress and deformation in static finite element simulation; Step 2: Use static finite element simulation software to simulate the stress and deformation of the parameterized model of the robot gripper structure under different dimensions and load conditions, and obtain stress and deformation simulation data. Step 2 includes the following steps: Step 2.1: In the static finite element simulation software, perform hexahedral mesh generation on the parametric model of the robot gripper structure; Step 2.2: In the static finite element simulation software, loads F under different working conditions are applied to different dimensional parameters DH and DV to perform static simulation and obtain stress and deformation simulation data; Step 2.2 specifically includes: (1) Based on the parameterized model of the robot gripper structure after the hexahedral mesh division in step 2.1, a fixed constraint is added to the robot gripper structure to indicate that the upper end of the robot gripper is tightly connected to the gripper base; (2) The lower end of the robot gripper contacts the target object being gripped, and the gripping force is provided by the cylinder. The weight of the target object and the gripping force of the cylinder are decomposed into two contact surfaces using the trigonometric function decomposition method, that is: ； wherein, is the holding force of the cylinder, m is the mass of the target object to be held, g is the acceleration of gravity, R is the angle of the clamping jaw, and n is the number of clamping arms. (3) Loads F under different working conditions are applied sequentially to DH and DV with different dimensional parameters. A and F B Static simulation was performed using ANSYS Workbench software to obtain stress and deformation simulation data. Step 3: Based on the obtained stress and deformation simulation data, construct a stress and deformation prediction model for the robot gripper containing the structural size parameters of the robot gripper and under different load conditions using a Kriging surrogate model; Step 4: By establishing a particle swarm optimization algorithm, the dimensional parameters that meet the stress and deformation constraints under actual load conditions are optimized. The optimal values ​​of the dimensional parameters are then substituted into the stress and deformation prediction model to obtain the dimensional parameters under actual working conditions and the stress and deformation under load, thereby optimizing the structural parameters of the robot gripper. The particle swarm optimization algorithm is established as follows: The mass of the robot gripper structure is used as the fitness function of the particle swarm optimization algorithm; The fitness function of the particle swarm optimization algorithm is: When the mass M of the beam reaches its minimum, that is: ； The constraints are set as follows: ； wherein D1, D2 are the optimal solution of a set of design variables; D1 max is the maximum value of D1, D2 min is the minimum value of D2, is the material allowable stress, DH is the wall thickness of the robot gripper structure, and DV is the wall spacing of the robot gripper structure. The maximum velocity V of the particles max 10% to 20% of the corresponding search space, set the inertia weight of the particle swarm algorithm value of the self-learning factor C1 and the value of setting the social learning factor C2, complete the construction of the particle swarm algorithm.

2. The method of claim 1, wherein, Step 1 includes the following steps: Step 1.1: Construct a 3D model of the robot gripper structure using 3D modeling software; Step 1.2: Input the wall thickness DH and wall spacing DV of the robot gripper structure into the 3D model of the robot gripper structure to obtain the parametric model of the robot gripper structure.

3. The method of claim 1, wherein, Step 3 includes the following steps: Step 3.1: Sample the stress and deformation simulation data obtained in Step 2 using Latin hypercube sampling; Step 3.2: Construct a dataset using all the sampled stress and deformation simulation data obtained in Step 3.1, and establish a Kriging surrogate calculation model for the robot gripper containing the structural size parameters of the robot gripper and different load conditions, i.e., a stress and deformation prediction model.

4. The method of claim 3, wherein, Step 3.1 specifically includes: The variable space is divided into n non-overlapping sub-intervals with equal probability. To ensure the uniformity of the sample points, a random and independent sample is taken in each sub-interval.

5. The method of claim 3, wherein, Step 3.2 specifically includes: The sampled stress and deformation simulation data are used to construct a dataset. The sample data after Latin hypercube sampling is substituted into the Kriging model that has been established in MATLAB to generate a deformation and stress prediction model. Using different dimensional parameters and different loads as inputs to the deformation and stress prediction model, and different stresses and deformations as outputs, the resulting deformation and stress prediction model is used as a Kriging proxy calculation model. After the Kriging surrogate model is established, the deformation and stress of the gripper component under different dimensional parameters DH, DV and different loads F are predicted using the Kriging surrogate model.

6. The method of claim 1, wherein, Step 4 includes the following steps: Step 4.1: Establish a particle swarm optimization algorithm. Determine the load based on the actual working conditions and use it as the input parameter for the particle swarm optimization algorithm; use different size parameters as particles. Step 4.2: Particle swarm initialization, i.e.: set the population size N; set the number of iterations to M, perform particle swarm optimization to obtain multiple individual optimal values ​​and a population optimal value; Step 4.3: Substitute the particles corresponding to the optimal values ​​obtained in Step 4.2 into the Kriging proxy calculation model to obtain the dimensional parameters of the actual working conditions and the stress and deformation under the load; Step 4.4: Determine whether the actual working condition dimensional parameters and stress and deformation under load obtained in Step 4.3 meet the set constraints. If they do, optimize the robot gripper structure parameters. If they do not, repeat Steps 1 to 4.

Images