CN111597645A - Optimal robot deburring process parameter selection method and system - Google Patents
Optimal robot deburring process parameter selection method and system Download PDFInfo
- Publication number
- CN111597645A CN111597645A CN202010262152.4A CN202010262152A CN111597645A CN 111597645 A CN111597645 A CN 111597645A CN 202010262152 A CN202010262152 A CN 202010262152A CN 111597645 A CN111597645 A CN 111597645A
- Authority
- CN
- China
- Prior art keywords
- population
- individuals
- model
- optimal
- chamfer width
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/17—Mechanical parametric or variational design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/04—Constraint-based CAD
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/06—Multi-objective optimisation, e.g. Pareto optimisation using simulated annealing [SA], ant colony algorithms or genetic algorithms [GA]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/28—Fuselage, exterior or interior
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Geometry (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Computer Hardware Design (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Analysis (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Computational Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Numerical Control (AREA)
- Manipulator (AREA)
Abstract
The invention discloses a robot deburring optimal process parameter selection method and system, wherein the rotation speed S, the feeding speed F and the floating force F of a floating main shaft are used as experimental variables, an experimental scheme is designed, a data set is constructed by combining the chamfer width L and the roughness Ra measured after a workpiece is deburred, the training of a multi-input and multi-output Gaussian process regression model is completed by adopting a K-fold cross validation method, all individuals of a population POP are predicted by utilizing the trained Gaussian process regression model, and an optimal design model is solved based on a multi-objective optimization algorithm NSGA-II, so that an optimal deburring process parameter combination meeting the chamfer width range is obtained. According to the invention, through a small number of experimental samples, the optimal deburring process parameter combination which can reach the target conditions of the maximum feeding speed and the minimum roughness within the required chamfer width range is quickly selected, and the deburring processing quality and efficiency are favorably improved.
Description
Technical Field
The invention relates to the technical field of robot machining, in particular to a method and a system for selecting optimal technological parameters for deburring of a robot floating spindle.
Background
Industrial robots are commonly used in the fields of deburring, welding, spraying and the like. Because burr shape differs, the size is less etc. reason, the unsteady main shaft that the industry often adopted air drive removes the burr, and the main shaft that floats has the compliance of a certain direction, can compensate the error that work piece positioning error, shape error etc. brought. However, the floating spindle driven by air cannot know the corresponding working rotating speed and floating force under a certain air pressure, and the mathematical relation of technological parameters to the deburring effect cannot be established through accurate numerical calculation in the deburring processing, so that the deburring processing quality and efficiency cannot be accurately controlled.
In the traditional deburring process parameter selection, a large number of experimental schemes are designed manually, and a large number of experimental data are obtained to analyze the influence trend of the process parameters so as to guide the machining. When the range of the deburring process parameters is wide and the inequality constraint conditions are required to be met, the traditional method cannot quickly and accurately find the optimal process parameter combination meeting the constraint conditions. In order to improve the machining quality of robot deburring and reduce the time required for determining technological parameters so as to meet industrial requirements, the rapid optimization and selection of the deburring technological parameters are necessary.
The deburring process parameter selection problem is a typical multi-objective optimization problem. In such problems, it is not possible to directly select a set of process parameter combinations such that all objective functions reach a maximum or minimum simultaneously. A common method for processing the multi-target optimization problem with the constraint condition is to convert the multi-target problem into a single-target problem by adopting a weight coefficient or penalty function form and solve the problem.
Disclosure of Invention
The invention aims to solve the technical problem that deburring process parameters cannot be determined quickly and accurately in the prior art, and provides a method and a system for selecting optimal deburring process parameters of a robot floating spindle.
The technical scheme adopted by the invention for solving the technical problems is as follows:
the optimal technological parameter selection method for robot deburring comprises the following steps:
s1, designing an experimental scheme by taking the rotating speed S, the feeding speed F and the floating force F of the floating main shaft as experimental variables, and constructing a data set by combining the chamfer width L and the roughness Ra measured after the workpiece is deburred;
s2, taking the rotation speed S, the feeding speed F and the floating force F of the data set floating main shaft as model input, taking the chamfer width L and the roughness Ra as model output, and finishing the training of a multiple-input multiple-output Gaussian process regression model by adopting a K-fold cross validation method;
s3, determining the upper limit and the lower limit of a decision vector, initializing a population in the decision vector range, predicting all individuals of the population POP by using a trained Gaussian process regression model, and outputting the chamfer width L and the roughness Ra corresponding to each individual;
s4, selecting a target function and determining constraint conditions, establishing an optimal design model of deburring process parameters, and solving the optimal design model based on a multi-objective optimization algorithm NSGA-II to obtain the optimal deburring process parameter combination meeting the chamfer width range.
In the above technical solution, the design experiment solution in step S1 is a hybrid design based on latin hypercube sampling and orthogonal experiment under the maximum and minimum criteria.
According to the above technical scheme, the training of the multiple-input multiple-output gaussian process regression model is completed by adopting a K-fold cross validation method in the step S2, and the method comprises the following steps:
s21, randomly dividing the data set constructed in the step S1 into K parts, selecting K-1 parts of data sets as training data sets each time, and using the remaining 1 part of data sets as testing data sets to finish K-fold data division;
s22, constructing a kernel function of a regression model of the Gaussian process by combining covariance functions of the Gaussian process, wherein the combination mode is the sum of square exponential covariance functions and linear covariance functions of the Gaussian process;
s23, solving the minimized negative log-likelihood function by adopting a K-fold cross validation method and combining a training data set and a conjugate gradient method to obtain a hyper-parameter, establishing K Gaussian process regression model sets, predicting a test data set by using the K Gaussian process regression models, calculating the mean square error of a model corresponding to each Gaussian process regression model, and selecting the Gaussian regression model MOD with the minimum mean square error of the modeliI is more than or equal to 1 and less than or equal to K, and the mean square error of the corresponding model is MSEi；
S24, setting a mean square error threshold MSE _ Min, if the mean square error MSE of the minimum model in the K Gaussian process regression models obtained in the step S23iIf the mean square error is larger than the mean square error threshold MSE _ Min, the step S21 is skipped, otherwise the minimum mean square error MSE is outputiCorresponding gaussian regression model MODiAnd (4) completing the training of the regression model of the Gaussian process.
In connection with the above technical solution, the individuals of the population POP in step S3 include a decision vector, a target vector and a category vector, the decision vector is composed of a rotation speed S, a feeding speed F and a floating force F, the target vector is composed of a feeding speed F and a roughness Ra, and the category vector is composed of a chamfer width L.
According to the above technical solution, the objective function in step S4 is the maximum feeding speed and the minimum roughness, and the constraint condition is the chamfer width working range [ L [ ]min,Lmax]。
According to the technical scheme, the step S4 of solving the optimization design model based on the multi-objective optimization algorithm NSGA-II comprises the following steps:
s41, performing rapid non-dominated sorting and congestion degree calculation on individuals in the population POP according to constraint conditions;
s42, selecting individuals in the POP which finishes the calculation of the ranking level and the crowdedness to generate a parent population, generating a child population through the crossing and variation of the parent population, combining the individuals in the parent population and the child population, performing rapid non-dominated ranking and crowdedness calculation on the individuals in the combined population, selecting the individuals with the top ranking level to form an elite population, and finishing one-time iterative optimization;
and S43, adding 1 to the iteration number, if the iteration number is smaller than a set population iteration threshold, predicting the individuals in the elite population in the step S42 by using a trained Gaussian process regression model, and then jumping to the step S41 to perform a new round of optimization, otherwise, ending the iteration optimization and outputting all the individuals in the elite population as an optimal process parameter combination.
In step S41, the fast non-dominated sorting and congestion degree calculation of the individuals in the population POP according to the constraint condition means:
firstly, classifying individuals in a POP population, and if the chamfer width L of the individuals in the POP population is within the chamfer width working range [ L ] defined by the constraint conditionmin,Lmax]If not, the individual is classified into an ideal population, otherwise, the individual is classified into a non-ideal population;
secondly, fast non-dominant sorting and congestion degree calculation are carried out on the individuals in the ideal population, the sorting level and the congestion degree of the individuals in the non-ideal population are directly set, wherein the sorting level is set to be N + i, N is the total number of the individuals in the population, i is the serial number of a certain individual in the population, i is more than or equal to 1 and less than or equal to N, and the congestion degree is set to be + ∞.
The optimal technological parameter selection system for robot deburring is provided, and is characterized by comprising the following components:
the experimental data set construction module is used for designing an experimental scheme by taking the rotating speed S, the feeding speed F and the floating force F of the floating main shaft as experimental variables, and constructing a data set by combining the chamfer width L and the roughness Ra measured after the workpiece is deburred;
the model training module is used for finishing the training of a multiple-input multiple-output Gaussian process regression model by taking the rotating speed S, the feeding speed F and the floating force F of the floating main shaft in the data set as model input, taking the chamfer width L and the roughness Ra as model output and adopting a K-fold cross validation method;
the individual prediction module is used for determining the upper limit and the lower limit of a decision vector, initializing a population in the decision vector range, predicting all individuals of the population POP by using a trained Gaussian process regression model, and outputting the chamfer width L and the roughness Ra corresponding to each individual;
and the iteration optimization module is used for selecting a target function and determining constraint conditions, establishing an optimal design model of deburring process parameters, and solving the optimal design model based on a multi-objective optimization algorithm NSGA-II to obtain an optimal deburring process parameter combination meeting the chamfer width range.
There is provided a computer storage medium having stored therein a computer program executable by a processor, the computer program performing the method for selecting optimal process parameters for robotic deburring as described in any one of the preceding claims.
The invention has the following beneficial effects: the invention provides a robot deburring optimal process parameter selection method and system, which are characterized in that the rotation speed S, the feeding speed F and the floating force F of a floating main shaft are used as experimental variables, an experimental scheme is designed, a data set is constructed by combining the chamfer width L and the roughness Ra measured after a workpiece is deburred, the training of a multi-input and multi-output Gaussian process regression model is completed by adopting a K-fold cross validation method, all individuals of population POP are predicted by utilizing the trained Gaussian process regression model, and an optimal design model is solved based on a multi-objective optimization algorithm NSGA-II, so that the optimal deburring process parameter combination meeting the chamfer width range is obtained. According to the invention, through a small number of experimental samples, the optimal deburring process parameter combination which can reach the target conditions of the maximum feeding speed and the minimum roughness within the required chamfer width range is quickly selected, and the deburring processing quality and efficiency are favorably improved.
Drawings
The invention will be further described with reference to the accompanying drawings and examples, in which:
FIG. 1 is a flow chart of the overall implementation of the method of the present invention;
FIG. 2 is a flow chart of the Gaussian process regression model training of the method of the present invention;
FIG. 3 is a flow chart of a method of the present invention for solving an optimal design model;
fig. 4 is an objective function solution set corresponding to the optimal deburring process parameter solved by the embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
As shown in fig. 1, the invention provides an optimal process parameter selection method for robot deburring, which comprises the following steps:
s1, designing an experimental scheme by taking the rotating speed S, the feeding speed F and the floating force F of the floating main shaft as experimental variables, and constructing a data set by combining the chamfer width L and the roughness Ra measured after the workpiece is deburred; as a specific example, the robot is KUKA-KR210R2700, the floating spindle is ATI axial floating spindle, the floating force of the spindle is measured by using a dynamometer Kistler9955A, and the chamfer width L and the roughness Ra of the deburred workpiece are measured by using a ZEISS laser confocal microscope; 30 sets of Latin hypercube experiments under the maximum and minimum criteria and 25 sets of orthogonal experiments are designed, and 55 experimental samples are collected.
And S2, taking the rotation speed S, the feeding speed F and the floating force F of the data set floating main shaft as model inputs, taking the chamfer width L and the roughness Ra as model outputs, and finishing the training of the multiple-input multiple-output Gaussian process regression model by adopting a K-fold cross validation method. The K-fold cross validation method is beneficial to improving the generalization ability of the regression model and improving the prediction precision of the process parameters by training the Gaussian process regression model.
S3, determining the upper limit and the lower limit of the decision vector, initializing the population in the decision vector range, predicting all individuals of the population POP by using the trained Gaussian process regression model, and outputting the chamfer width L and the roughness Ra corresponding to each individual.
S4, selecting a target function and determining constraint conditions, establishing an optimal design model of deburring process parameters, and solving the optimal design model based on a multi-objective optimization algorithm NSGA-II to obtain the optimal deburring process parameter combination meeting the chamfer width range.
According to the method, the optimal deburring process parameter combination which achieves the target conditions of the maximum feeding speed and the minimum roughness within the required chamfer width range is quickly selected through a small number of experimental samples, so that the deburring processing quality and efficiency are improved, and the time required for determining the process parameters is shortened.
Further, the design experiment scheme in step S1 is a hybrid design of latin hypercube sampling and orthogonal experiment based on the maximum and minimum criteria, which effectively ensures space fullness of small sample data.
Further, as shown in fig. 2, the training of the multiple-input multiple-output gaussian process regression model is completed by using a K-fold cross validation method in step S2, which includes the following steps:
s21, randomly dividing the data set constructed in the step S1 into K parts, selecting K-1 parts of data sets as training data sets each time, and using the remaining 1 part of data sets as testing data sets to finish K-fold data division; as a specific example, the 55 collected experimental samples were divided into 5 samples, and 11 samples were collected.
S22, constructing a kernel function of the regression model of the Gaussian process by combining the covariance functions of the Gaussian process, wherein the combination mode is the sum of the square exponential covariance function and the linear covariance function of the Gaussian process, and the kernel function constructed by the method has a better generalization function.
S23, solving the minimized negative log-likelihood function by adopting a K-fold cross validation method and combining a training data set and a conjugate gradient method to obtain hyper-parameters, establishing K Gaussian process regression model sets, predicting the test data set by using the K Gaussian process regression models, and calculating each setThe mean square error of the model corresponding to the regression model of the Gaussian process is selected, and the Gaussian regression model MOD with the minimum mean square error of the model is selectediI is more than or equal to 1 and less than or equal to K, and the mean square error of the corresponding model is MSEi。
S24, setting a mean square error threshold MSE _ Min, if the mean square error MSE of the minimum model in the K Gaussian process regression models obtained in the step S23iIf the mean square error is larger than the mean square error threshold MSE _ Min, the step S21 is skipped, otherwise the minimum mean square error MSE is outputiCorresponding gaussian regression model MODiAnd (4) completing the training of the regression model of the Gaussian process.
The Gaussian process regression is a machine learning algorithm combining Bayesian inference and statistical learning theory, and has advantages and good generalization capability in processing complex problems of small samples, high dimensionality, nonlinearity and the like. The essence of the Gaussian process regression is that probability modeling is used for output prediction, and compared with a neural network and a support vector machine, the Gaussian process regression has the characteristics of easiness in implementation, strict significance of output probability and the like.
By adopting a K-fold cross validation method, the trained Gaussian process regression model repeatedly iterates to find out the optimal hyperparameter through continuous cross validation of a small number of experimental samples, thereby being beneficial to improving the generalization capability of the regression model and improving the prediction precision of the process parameters.
Further, the individuals of the population POP in the step S3 include a decision vector, a target vector and a category vector, the decision vector is composed of a rotation speed S, a feeding speed F and a floating force F, the upper and lower limits of the decision vector are determined by the normal working range of the performance parameters of the floating spindle, the target vector is composed of the feeding speed F and the roughness Ra, and the category vector is composed of the chamfer width L.
Further, as shown in fig. 3, the objective function in step S4 is the maximum feed speed and the minimum roughness, and the constraint condition is the chamfer width working range [ L [ ]min,Lmax]。
As a specific embodiment, the normal working range of the floating main shaft is that the rotating speed S is more than or equal to 1000r/min and less than or equal to 5000r/min, the feeding speed F is more than or equal to 1000mm/min and less than or equal to 3000mm/min, and the floating force F is more than or equal to 10N and less than or equal to 60N.
According to the technical scheme, the step S4 of solving the optimization design model based on the multi-objective optimization algorithm NSGA-II comprises the following steps:
s41, performing rapid non-dominated sorting and congestion degree calculation on individuals in the population POP according to constraint conditions;
s42, selecting individuals in the POP which finishes the calculation of the ranking level and the crowdedness to generate a parent population, generating a child population through the crossing and variation of the parent population, combining the individuals in the parent population and the child population, performing rapid non-dominated ranking and crowdedness calculation on the individuals in the combined population, selecting the individuals with the top ranking level to form an elite population, and finishing one-time iterative optimization;
and S43, adding 1 to the iteration number, if the iteration number is smaller than a set population iteration threshold, predicting the individuals in the elite population in the step S42 by using a trained Gaussian process regression model, and then jumping to the step S41 to perform a new round of optimization, otherwise, ending the iteration optimization and outputting all the individuals in the elite population as an optimal process parameter combination.
As a specific embodiment, in order to ensure that an approximately optimal solution is obtained, the threshold of the number of population iterations is set to 500.
Further, the fast non-dominated sorting and congestion degree calculation of the individuals in the population POP according to the constraint conditions in step S41 means:
firstly, classifying individuals in a POP population, and if the chamfer width L of the individuals in the POP population is within the chamfer width working range [ L ] defined by the constraint conditionmin,Lmax]If not, the individual is classified into an ideal population, otherwise, the individual is classified into a non-ideal population;
secondly, fast non-dominant sorting and crowding calculation are carried out on the individuals in the ideal population, the sorting level and the crowding degree of the individuals in the non-ideal population are directly set, wherein the sorting level is set to be N + i, N is the total number of the individuals in the population, i is the serial number of a certain individual in the population, i is more than or equal to 1 and less than or equal to N, and the crowding degree is set to be + ∞.
The optimal technological parameter selection system for robot deburring is provided, and comprises:
the experimental data set construction module is used for designing an experimental scheme by taking the rotating speed S, the feeding speed F and the floating force F of the floating main shaft as experimental variables, and constructing a data set by combining the chamfer width L and the roughness Ra measured after the workpiece is deburred;
the model training module is used for finishing the training of a multiple-input multiple-output Gaussian process regression model by taking the rotating speed S, the feeding speed F and the floating force F of the floating main shaft in the data set as model input, taking the chamfer width L and the roughness Ra as model output and adopting a K-fold cross validation method;
the individual prediction module is used for determining the upper limit and the lower limit of a decision vector, initializing a population in the decision vector range, predicting all individuals of the population POP by using a trained Gaussian process regression model, and outputting the chamfer width L and the roughness Ra corresponding to each individual;
and the iteration optimization module is used for selecting a target function and determining constraint conditions, establishing an optimal design model of deburring process parameters, and solving the optimal design model based on a multi-objective optimization algorithm NSGA-II to obtain an optimal deburring process parameter combination meeting the chamfer width range.
There is provided a computer storage medium having stored therein a computer program executable by a processor, the computer program performing the method for selecting optimal process parameters for robotic deburring as described in any one of the preceding claims.
The K-fold cross validation method is adopted to train the Gaussian process regression model, so that the generalization capability of the regression model is improved, the prediction precision of the process parameters is improved, the optimal deburring process parameter combination which meets the target conditions of the maximum feeding speed and the minimum roughness in the required chamfer width range can be quickly selected by solving the optimization design model based on the multi-objective optimization algorithm NSGA-II, and the deburring processing quality and efficiency are improved. As a specific embodiment, the solution set of the objective function corresponding to the solved optimal deburring process parameter is shown in fig. 4.
It will be understood that modifications and variations can be made by persons skilled in the art in light of the above teachings and all such modifications and variations are intended to be included within the scope of the invention as defined in the appended claims.
Claims (9)
1. The optimal technological parameter selection method for robot deburring is characterized by comprising the following steps:
s1, designing an experimental scheme by taking the rotating speed S, the feeding speed F and the floating force F of the floating main shaft as experimental variables, and constructing a data set by combining the chamfer width L and the roughness Ra measured after the workpiece is deburred;
s2, taking the rotation speed S, the feeding speed F and the floating force F of the data set floating main shaft as model input, taking the chamfer width L and the roughness Ra as model output, and finishing the training of a multiple-input multiple-output Gaussian process regression model by adopting a K-fold cross validation method;
s3, determining the upper limit and the lower limit of a decision vector, initializing a population in the decision vector range, predicting all individuals of the population POP by using a trained Gaussian process regression model, and outputting the chamfer width L and the roughness Ra corresponding to each individual;
s4, selecting a target function and determining constraint conditions, establishing an optimal design model of deburring process parameters, and solving the optimal design model based on a multi-objective optimization algorithm NSGA-II to obtain the optimal deburring process parameter combination meeting the chamfer width range.
2. The method of claim 1, wherein the design experiment plan in step S1 is a hybrid design of latin hypercube sampling and orthogonal experiments based on the maximum and minimum criteria.
3. The method of claim 1, wherein the training of the multiple-input multiple-output gaussian process regression model is completed by using a K-fold cross validation method in step S2, and the method comprises the following steps:
s21, randomly dividing the data set constructed in the step S1 into K parts, selecting K-1 parts of data sets as training data sets each time, and using the remaining 1 part of data sets as testing data sets to finish K-fold data division;
s22, constructing a kernel function of a regression model of the Gaussian process by combining covariance functions of the Gaussian process, wherein the combination mode is the sum of square exponential covariance functions and linear covariance functions of the Gaussian process;
s23, solving the minimized negative log-likelihood function by adopting a K-fold cross validation method and combining a training data set and a conjugate gradient method to obtain a hyper-parameter, establishing K Gaussian process regression model sets, predicting a test data set by using the K Gaussian process regression models, calculating the mean square error of a model corresponding to each Gaussian process regression model, and selecting the Gaussian regression model MOD with the minimum mean square error of the modeliI is more than or equal to 1 and less than or equal to K, and the mean square error of the corresponding model is MSEi；
S24, setting a mean square error threshold MSE _ Min, if the mean square error MSE of the minimum model in the K Gaussian process regression models obtained in the step S23iIf the mean square error is larger than the mean square error threshold MSE _ Min, the step S21 is skipped, otherwise the minimum mean square error MSE is outputiCorresponding gaussian regression model MODiAnd (4) completing the training of the regression model of the Gaussian process.
4. The method according to claim 1, characterized in that the individuals of the population POP in step S3 include a decision vector consisting of a rotation speed S, a feed speed F, a floating force F, a target vector consisting of a feed speed F, a roughness Ra, and a category vector consisting of a chamfer width L.
5. The method according to claim 1, wherein the objective function in step S4 is maximum feed speed and minimum roughness, and the constraint condition is the chamfer width working range [ L [ ]min,Lmax]。
6. The method according to claim 1, wherein the step of solving the optimal design model based on the multi-objective optimization algorithm NSGA-II in the step S4 comprises the following steps:
s41, performing rapid non-dominated sorting and congestion degree calculation on individuals in the population POP according to constraint conditions;
s42, selecting individuals in the POP which finishes the calculation of the ranking level and the crowdedness to generate a parent population, generating a child population through the crossing and variation of the parent population, combining the individuals in the parent population and the child population, performing rapid non-dominated ranking and crowdedness calculation on the individuals in the combined population, selecting the individuals with the top ranking level to form an elite population, and finishing one-time iterative optimization;
and S43, adding 1 to the iteration number, if the iteration number is smaller than a set population iteration threshold, predicting the individuals in the elite population in the step S42 by using a trained Gaussian process regression model, and then jumping to the step S41 to perform a new round of optimization, otherwise, ending the iteration optimization and outputting all the individuals in the elite population as an optimal process parameter combination.
7. The method according to claim 5 or 6, wherein the fast non-dominated sorting and congestion degree calculation of the individuals in the population POP according to the constraint conditions in step S41 includes:
firstly, classifying individuals in a POP population, and if the chamfer width L of the individuals in the POP population is within the chamfer width working range [ L ] defined by the constraint conditionmin,Lmax]If not, the individual is classified into an ideal population, otherwise, the individual is classified into a non-ideal population;
secondly, fast non-dominant sorting and congestion degree calculation are carried out on the individuals in the ideal population, the sorting level and the congestion degree of the individuals in the non-ideal population are directly set, wherein the sorting level is set to be N + i, N is the total number of the individuals in the population, i is the serial number of a certain individual in the population, i is more than or equal to 1 and less than or equal to N, and the congestion degree is set to be + ∞.
8. The optimal technological parameter selection system for robot deburring is characterized by comprising the following components:
the experimental data set construction module is used for designing an experimental scheme by taking the rotating speed S, the feeding speed F and the floating force F of the floating main shaft as experimental variables, and constructing a data set by combining the chamfer width L and the roughness Ra measured after the workpiece is deburred;
the model training module is used for finishing the training of a multiple-input multiple-output Gaussian process regression model by taking the rotating speed S, the feeding speed F and the floating force F of the floating main shaft in the data set as model input, taking the chamfer width L and the roughness Ra as model output and adopting a K-fold cross validation method;
the individual prediction module is used for determining the upper limit and the lower limit of a decision vector, initializing a population in the decision vector range, predicting all individuals of the population POP by using a trained Gaussian process regression model, and outputting the chamfer width L and the roughness Ra corresponding to each individual;
and the iteration optimization module is used for selecting a target function and determining constraint conditions, establishing an optimal design model of deburring process parameters, and solving the optimal design model based on a multi-objective optimization algorithm NSGA-II to obtain an optimal deburring process parameter combination meeting the chamfer width range.
9. A computer storage medium, characterized in that a computer program executable by a processor is stored therein, the computer program executing the robot deburring optimal process parameter selection method as claimed in any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010262152.4A CN111597645B (en) | 2020-04-06 | 2020-04-06 | Robot deburring optimal process parameter selection method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010262152.4A CN111597645B (en) | 2020-04-06 | 2020-04-06 | Robot deburring optimal process parameter selection method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111597645A true CN111597645A (en) | 2020-08-28 |
| CN111597645B CN111597645B (en) | 2023-07-18 |
Family
ID=72185165
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010262152.4A Active CN111597645B (en) | 2020-04-06 | 2020-04-06 | Robot deburring optimal process parameter selection method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111597645B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112580682A (en) * | 2020-11-16 | 2021-03-30 | 河海大学 | Dam monitoring system missing data reconstruction method based on multitask Gaussian process |
| CN117331355A (en) * | 2023-09-27 | 2024-01-02 | 江苏中科云控智能工业装备有限公司 | Die casting deburring process regulation and control system and method based on neural network |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2821872A1 (en) * | 2013-07-01 | 2015-01-07 | Getting Robotika S.L. | Robotised cell, installation and method for deburring parts |
| CN108733912A (en) * | 2018-05-17 | 2018-11-02 | 南京理工大学 | A kind of robot rotary ultrasonic drilling CFRP/ aluminium alloy laminated construction outlet burr height prediction technique |
| DE102019001783A1 (en) * | 2018-03-20 | 2019-09-26 | Fanuc Corporation | CONTROL, MACHINE LEARNING DEVICE AND SYSTEM |
| DE102019002156A1 (en) * | 2018-04-02 | 2019-10-02 | Fanuc Corporation | Controller and machine learning device |
-
2020
- 2020-04-06 CN CN202010262152.4A patent/CN111597645B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2821872A1 (en) * | 2013-07-01 | 2015-01-07 | Getting Robotika S.L. | Robotised cell, installation and method for deburring parts |
| DE102019001783A1 (en) * | 2018-03-20 | 2019-09-26 | Fanuc Corporation | CONTROL, MACHINE LEARNING DEVICE AND SYSTEM |
| DE102019002156A1 (en) * | 2018-04-02 | 2019-10-02 | Fanuc Corporation | Controller and machine learning device |
| CN108733912A (en) * | 2018-05-17 | 2018-11-02 | 南京理工大学 | A kind of robot rotary ultrasonic drilling CFRP/ aluminium alloy laminated construction outlet burr height prediction technique |
Non-Patent Citations (4)
| Title |
|---|
| ARINDAM DEBROY, SHANKAR CHAKRABORTY: "Non-conventional optimization techniques in optimizing non-traditional machining processes: A review" * |
| F. LEO PRINCELY,ETC: "Application of TOPSIS method for optimization of process parameters in robotic deburring" * |
| 李波,ETC: "基于改进遗传算法的机器人去毛刺工艺参数优化" * |
| 王连生: "基于机器视觉的工业机器人去毛刺加工效率优化研究" * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112580682A (en) * | 2020-11-16 | 2021-03-30 | 河海大学 | Dam monitoring system missing data reconstruction method based on multitask Gaussian process |
| CN117331355A (en) * | 2023-09-27 | 2024-01-02 | 江苏中科云控智能工业装备有限公司 | Die casting deburring process regulation and control system and method based on neural network |
| CN117331355B (en) * | 2023-09-27 | 2024-04-23 | 江苏中科云控智能工业装备有限公司 | Die casting deburring process regulation and control system and method based on neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111597645B (en) | 2023-07-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109828532B (en) | GA-GBRT-based surface roughness prediction method and technological parameter optimization method | |
| Abo-Hammour et al. | An optimization algorithm for solving systems of singular boundary value problems | |
| Razfar et al. | Optimum surface roughness prediction in face milling by using neural network and harmony search algorithm | |
| Tesch et al. | Using response surfaces and expected improvement to optimize snake robot gait parameters | |
| CN111797469B (en) | Aeroengine case technological parameter optimization method based on machining cutter relieving deformation constraint | |
| CN111597645A (en) | Optimal robot deburring process parameter selection method and system | |
| CN110363214B (en) | GWA-SVM-based contact state recognition method for robot assembly | |
| Cao et al. | A novel multi-objective optimization approach of machining parameters with small sample problem in gear hobbing | |
| Roy et al. | Disruption in the US machine tool industry: the role of inhouse users and pre-disruption component experience in firm response | |
| CN114492198A (en) | Cutting force prediction method based on improved PSO algorithm assisted SVM algorithm | |
| CN115640898A (en) | Large-scale flexible job shop scheduling method based on DDQN algorithm | |
| CN113420508A (en) | Unit combination calculation method based on LSTM | |
| GB2617741A (en) | Multi-level multi-objective automated machine learning | |
| CN107844658A (en) | The steering wheel trend prediction method that Method Using Relevance Vector Machine merges with autoregressive moving average | |
| Rad et al. | GP-RVM: Genetic programing-based symbolic regression using relevance vector machine | |
| CN116533063A (en) | Online prediction method for residual life of numerically controlled lathe tool | |
| Sharma et al. | Analysis of software effort estimation based on Story Point and lines of code using machine learning | |
| Chen et al. | Learning to plan via neural exploration-exploitation trees | |
| Greis et al. | Stability modeling for chatter avoidance in self-aware machining: an application of physics-guided machine learning | |
| Wei et al. | A bayesian approach to robust inverse reinforcement learning | |
| Yoon et al. | Learning to improve performance during non-repetitive tasks performed by robots | |
| Osman | Optimizing Model Base Predictive Control for Combustion Boiler Process at High Model Uncertainty | |
| Ahmed et al. | Towards futuristic autonomous experimentation--A surprise-Reacting sequential experiment policy | |
| Iba et al. | GP-RVM: Genetic programing-based symbolic regression using relevance vector machine | |
| Zhang et al. | Milling force prediction of titanium alloy based on support vector machine and ant colony optimization |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |