CN110674598A - Injection molding process optimization method based on support vector machine and particle swarm optimization - Google Patents
Injection molding process optimization method based on support vector machine and particle swarm optimization Download PDFInfo
- Publication number
- CN110674598A CN110674598A CN201910789733.0A CN201910789733A CN110674598A CN 110674598 A CN110674598 A CN 110674598A CN 201910789733 A CN201910789733 A CN 201910789733A CN 110674598 A CN110674598 A CN 110674598A
- Authority
- CN
- China
- Prior art keywords
- support vector
- vector machine
- injection molding
- particle swarm
- model
- 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.)
- Pending
Links
- 238000012706 support-vector machine Methods 0.000 title claims abstract description 97
- 238000000034 method Methods 0.000 title claims abstract description 81
- 238000005457 optimization Methods 0.000 title claims abstract description 54
- 239000002245 particle Substances 0.000 title claims abstract description 53
- 238000001746 injection moulding Methods 0.000 title claims abstract description 42
- 238000012360 testing method Methods 0.000 claims abstract description 71
- 230000008569 process Effects 0.000 claims abstract description 52
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 38
- 238000004519 manufacturing process Methods 0.000 claims abstract description 18
- 238000012549 training Methods 0.000 claims abstract description 18
- 238000002347 injection Methods 0.000 claims abstract description 15
- 239000007924 injection Substances 0.000 claims abstract description 15
- 230000002068 genetic effect Effects 0.000 claims abstract description 14
- 238000013461 design Methods 0.000 claims abstract description 12
- 238000013507 mapping Methods 0.000 claims abstract description 8
- 238000004458 analytical method Methods 0.000 claims description 15
- 238000000611 regression analysis Methods 0.000 claims description 11
- 230000004044 response Effects 0.000 claims description 8
- 238000002156 mixing Methods 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 5
- 238000012795 verification Methods 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 4
- 239000000463 material Substances 0.000 claims description 4
- 238000005094 computer simulation Methods 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 2
- 239000000047 product Substances 0.000 description 20
- 238000001816 cooling Methods 0.000 description 12
- 238000011160 research Methods 0.000 description 10
- 238000004088 simulation Methods 0.000 description 9
- 230000006870 function Effects 0.000 description 8
- 238000000465 moulding Methods 0.000 description 7
- 230000006872 improvement Effects 0.000 description 6
- 239000000243 solution Substances 0.000 description 5
- 230000008859 change Effects 0.000 description 4
- 239000000155 melt Substances 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 230000007547 defect Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000012417 linear regression Methods 0.000 description 3
- 238000013386 optimize process Methods 0.000 description 3
- 230000009471 action Effects 0.000 description 2
- 238000000540 analysis of variance Methods 0.000 description 2
- 238000000576 coating method Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 238000009827 uniform distribution Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- RNFJDJUURJAICM-UHFFFAOYSA-N 2,2,4,4,6,6-hexaphenoxy-1,3,5-triaza-2$l^{5},4$l^{5},6$l^{5}-triphosphacyclohexa-1,3,5-triene Chemical compound N=1P(OC=2C=CC=CC=2)(OC=2C=CC=CC=2)=NP(OC=2C=CC=CC=2)(OC=2C=CC=CC=2)=NP=1(OC=1C=CC=CC=1)OC1=CC=CC=C1 RNFJDJUURJAICM-UHFFFAOYSA-N 0.000 description 1
- OYPRJOBELJOOCE-UHFFFAOYSA-N Calcium Chemical compound [Ca] OYPRJOBELJOOCE-UHFFFAOYSA-N 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 239000006227 byproduct Substances 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 239000011248 coating agent Substances 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- KVFIJIWMDBAGDP-UHFFFAOYSA-N ethylpyrazine Chemical compound CCC1=CN=CC=N1 KVFIJIWMDBAGDP-UHFFFAOYSA-N 0.000 description 1
- 239000003063 flame retardant Substances 0.000 description 1
- 238000003052 fractional factorial design Methods 0.000 description 1
- 230000017525 heat dissipation Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 229920000642 polymer Polymers 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
Abstract
The invention discloses an injection molding process optimization method based on a support vector machine and a particle swarm algorithm, which is a method for optimizing Support Vector Machine (SVM) parameters by a Genetic Algorithm (GA) and optimizing optimal process parameters by combining the particle swarm algorithm (PSO). firstly, a process parameter is selected as an influence factor according to the specific production condition and quality index of an injection molding product to design a uniform test; secondly, dividing the uniform test data into training data and test data, optimizing parameters of a support vector machine model by using a genetic algorithm and constructing the support vector machine model to obtain a nonlinear mapping relation between injection molding process parameters and quality indexes; and finally, optimizing the process parameters by combining a particle swarm algorithm to obtain the optimal process parameters. The invention can effectively determine the optimal variable value and obtain the optimal process parameter when multiple variables are involved and the variable range difference is large, has high prediction precision and can effectively improve the production quality of injection products.
Description
Technical Field
The invention relates to a process optimization method, in particular to an injection molding process optimization method based on a support vector machine and a particle swarm algorithm, and belongs to the technical field of injection molding processing.
Background
The injection molding has the advantages of low cost and high efficiency, is widely applied to plastic part processing, and is an important manufacturing process in the global plastic industry. Injection molding involves three stages: and in the filling stage, the pressure maintaining stage and the cooling stage, molten plastic is injected into the mold cavity to form the required part. The main factors influencing the molding quality of the plastic part are a mold and an injection molding process, and the product quality can be improved by optimizing the mold and the injection molding process.
The cooling time is the key stage of the injection molding cycle, and accounts for more than half of the total time of the mold, and the good cooling system is beneficial to improving the product quality and the production efficiency. Clement et al studied cooling systems to effectively improve product quality and production efficiency. In addition, the surface of the cavity has great influence on the flow and the molding quality of the plastic melt, and Masato and the like research on the influence of different mold coatings on thin-wall injection molding. In addition, the relation between the process parameters and the quality is complex, the coupling is strong, a lot of researches are carried out aiming at the optimization of the process parameters in the industry at present, for example, Ginghtong and the like determine the optimal process parameters through orthogonal tests and linear regression analysis, Heidari and the like optimize the process parameters and effectively improve the product quality by combining fractional factorial design and variance analysis, Kitayama and the like determine the optimal process parameters by using sequence approximation optimization of a RBF network, Oliaei and the like research the processing performance of three polymers by using a neural network and variance analysis, Kitayama and the like research the relation between weld mark temperature and clamping force by using multi-objective optimization of the RBF network, Bendingh and the like predict the optimal process parameters of double-aspheric lens molding by using a mixed artificial neural network and a Particle Swarm Optimization (PSO), Li and the like optimize injection molding processes by combining a Taguchi method, a response surface method and a NSGA-II method, KC and the like determine the optimal injection molding parameters and important variables by using a Taguchi method and variance analysis, and the like.
However, the research on the mold cooling system and the cavity coating material can effectively improve the product quality, but the cost for improving the product quality relative to the optimization of the injection molding process parameters is high. In addition, when the research process parameters are optimized, the test points cannot be uniformly distributed due to different process parameter variation ranges, and the horizontal division span difference of each factor is large. In addition, because the number of tests is limited, the method (such as a neural network) based on the empirical risk minimization theory cannot meet the requirement of infinite data volume in use, so that the prediction precision is not high, and the complex relationship between the influencing factors and the quality index cannot be reflected to the maximum extent.
Disclosure of Invention
Aiming at the problems, the invention provides an injection molding process optimization method based on a support vector machine and a particle swarm algorithm, which can effectively determine the optimal variable value and obtain the optimal process parameter when multiple variables are involved and the variable range difference is large, has high prediction precision and can effectively improve the production quality of injection molding products.
In order to achieve the purpose, the injection molding process optimization method based on the support vector machine and the particle swarm optimization is a method for optimizing the parameters of the Support Vector Machine (SVM) by a Genetic Algorithm (GA) and optimizing the optimal process parameters by combining the Particle Swarm Optimization (PSO), and firstly, the process parameters are selected as influence factors according to the specific production condition and quality indexes of an injection molding product to design a uniform test; secondly, dividing the uniform test data into training data and test data, optimizing parameters of a support vector machine model by using a genetic algorithm and constructing the support vector machine model to obtain a nonlinear mapping relation between injection molding process parameters and quality indexes; and finally, optimizing the process parameters by combining a particle swarm algorithm to obtain the optimal process parameters.
As a further improvement scheme of the invention, in the design of the uniformity test, after determining each influence factor and an optimization target, horizontally dividing and designing a mixed horizontal uniformity test table according to the variation range of each influence factor, generating the mixed uniformity test table, simulating the warping amount of the plastic part at different levels by Moldflow software, then establishing a regression model for regression analysis, and obtaining the process parameter with the minimum warping amount of the plastic part as a verification index.
The mixing level uniformity test table is:
wherein n is the number of tests, t1，t2，t3Is the horizontal number of columns, q1，q2，q3Respectively, the horizontal number is t1，t2，t3The number of columns of (c).
As a further improvement of the present invention, a homogeneous mixing test table is generated using the DPS data system.
As a further improvement scheme of the invention, the regression model adopts a quadratic complete polynomial regression model:
where y is the optimization target, n is the number of variables, b0、bi、bii、bijAs a model parameter, xi、xjTo design variables, ε is the precision error.
As a further improvement scheme of the invention, when the regression model is established for regression analysis, variance analysis is carried out after all coefficients of the regression model are obtained, and the coefficient which can be determined is adjusted by calculation
The verification is carried out in such a way that,
in the formula, n is the number of samples, and k is the number of parameters in the fitting equation;
and solving the regression model within the selected process parameter range to obtain the process parameter with the minimum plastic part warping amount.
As a further improvement of the invention, the steps of constructing the support vector machine model are as follows:
step 1: determining various influencing factors and quality indexes according to the material and the actual production condition of the plastic part, and determining the variation range of the various influencing factors;
step 2: obtaining a response value of the quality index of the plastic part according to a uniform test and a computer simulation value;
and step 3: aiming at different units and variable ranges of all the influence factors, carrying out normalization processing on the influence factors and the quality index values to eliminate the influence of dimensions on analysis, and taking normalized data as a training sample and a test sample of a support vector machine model;
and 4, step 4: selecting a kernel function of the support vector machine model, and optimizing related parameters of the support vector machine model to obtain the optimal model parameter combination;
and 5: training a support vector machine model by using a training sample, and checking the precision of the support vector machine model;
step 6: and if the accuracy of the support vector machine model does not meet the requirement, increasing the number of training samples, turning to the step 2, and retraining the support vector machine model until an ideal support vector machine model is obtained.
As a further improvement scheme of the invention, the extreme value is searched for the support vector machine model by using the particle swarm optimization, and the trained support vector machine model is used as a fitness function to be optimized to obtain the optimal process parameters.
Compared with the prior art, the invention uses the uniform test, effectively reduces the test times when the horizontal number is more, solves the problem of uneven horizontal division caused by different multi-factor change ranges, and ensures that the test points are distributed more uniformly in the change range; by polynomial regression analysis, the relation between the influence factors and the response values is analyzed, the prediction precision is high, the plastic part warping deformation is effectively reduced, and the product production quality is improved; optimizing the model parameters of the support vector machine through a genetic algorithm, solving the problem of SVM model parameter selection, effectively improving the complex relationship between the mapping factors and the response values of the support vector machine and laying a foundation for optimizing process parameters; and searching an extreme value of the trained SVM model by using a particle swarm algorithm to obtain an optimal process parameter, inputting the optimal process parameter obtained by optimizing the particle swarm algorithm into the Moldflow for simulation, wherein the result shows that the quality index of the optimized product is obviously improved, and the optimized process parameter can effectively improve the quality of the product.
Drawings
FIG. 1 is a model diagram of a finite element analysis of a power adapter housing molding;
FIG. 2 is a graph of warpage simulation values obtained by inputting the process parameters of minimum warpage of plastic parts obtained by regression analysis of uniformity tests into Moldflow software;
FIG. 3 is a flow chart of SVM construction;
FIG. 4 is a flow chart of a genetic algorithm to optimize SVM parameters;
FIG. 5 is a diagram of an SVM model;
FIG. 6 is a SVM fit graph;
FIG. 7 is a graph of SVM prediction relative error;
FIG. 8 is a particle swarm optimization flow chart;
FIG. 9 is a diagram of a warping iterative optimization process;
FIG. 10 is a graph of optimized optimal process parameters input into the Moldflow software to obtain warp prediction values.
Detailed Description
The injection molding process optimization method based on the support vector machine and the particle swarm optimization is a method for optimizing the parameters of the Support Vector Machine (SVM) by a Genetic Algorithm (GA) and optimizing the optimal process parameters by combining the Particle Swarm Optimization (PSO), and aims to effectively determine the optimal variable value when multiple variables are involved and the variable range difference is large. Firstly, selecting process parameters as influencing factors to design a uniform test according to the specific production condition and quality index of an injection molding product; secondly, dividing the uniform test data into training data and test data, optimizing parameters of a support vector machine model by using a genetic algorithm and constructing the support vector machine model to obtain a nonlinear mapping relation between injection molding process parameters and quality indexes; and finally, optimizing the process parameters by combining a particle swarm algorithm to obtain the optimal process parameters.
The present invention will be further described below by taking a certain power adapter housing plastic as an example.
The power adapter is also called an external power supply, and is power supply conversion equipment for small-sized portable electronic equipment and electronic appliances, a power adapter shell is one of main parts of a power adapter product, an air inlet window grid and an exhaust fan for air convection heat dissipation are usually arranged on a shell-shaped structure of the power adapter, and the shell of the power adapter is mostly formed by plastic direct injection molding at present. A power adapter shell plastic part of a finite element analysis model shown in figure 1 is a thin-wall, small-volume, symmetrical and compact structure, the external dimension of the power adapter shell plastic part is 145mm multiplied by 130mm multiplied by 55mm, the wall thickness of the power adapter shell plastic part is 1-2 mm, the power adapter shell plastic part belongs to a typical thin-shell injection molding part, and the product is made of flame-retardant ABS and is marked by PA-765A.
Selection of influencing factors and optimization objectives
In the injection molding process, there are many factors that affect the quality of the product, such as: the method comprises the steps of selecting a mold, and selecting a mold, wherein the mold comprises a melt temperature, a mold temperature, a cooling time, an injection time, a pressure maintaining pressure, a pressure maintaining time and the like, different factors have different influences on injection molding, and a plastic part molding defect can be caused by combined action of a plurality of factors, so that related influencing factors need to be selected pertinently according to different optimization targets.
According to the defects generated during the production of the power adapter shell plastic part, the main defect is warp deformation, which causes the poor matching of the power adapter shell and other parts to generate gaps, so the warp deformation is used as a quality index.
According to the specific production condition of the power adapter shell plastic part, the selected influencing factor is the melt temperature (T)meltDEG C), mold temperature (T)moldDEG C), injection pressure (P)injMPa), injection time (t)injS), cooling time (t)cS), holding pressure (P)hMPa) and dwell time (t)hAnd s) 7 process parameters. The variation range of the process parameters is determined by the actual materials and productionThe circumference is as follows:
(II) design of homogeneous test
The homogeneous test has been proposed by Wangyuan and Fangkaitai and has been widely applied to the fields of manufacturing, pharmacy, science, engineering and the like. The uniform test design only considers the uniform distribution of the test points in the test range, and the test times can be greatly reduced.
After determining the influence factors and the optimization target, selecting a test design method according to test requirements, and selecting a uniform test as the test design method for considering the relative uniformity of horizontal division of the influence factors and reducing the test times as much as possible due to larger variation range span of related factors.
Because the variation range of each influence factor is greatly different, the horizontal division of each influence factor is different: the injection pressure and the holding pressure were classified into 60 levels, the melt temperature and the mold temperature were classified into 30 levels, the cooling time and the holding time were classified into 20 levels, and the injection time was classified into 4 levels. The general form of the mixing level homogeneity test table is:
wherein n is the number of tests, t1，t2，t3Is the horizontal number of columns, q1，q2，q3Respectively, the horizontal number is t1，t2，t3The number of columns of (c).
Because the number of times of the uniform test is more, the general uniform table is not suitable, and the DPS data system is utilized to generate the mixed uniform test table U60(602×302×202×41) And simulating the magnitude of the warping amount of the molded part of each influencing factor at different levels by Moldflow software, wherein the test scheme and the results are shown in Table 1.
Table 1 homogeneous test table and simulation results
The uniform test only considers the uniform distribution of test points in the test range and does not consider the uniformity, and common analysis methods are an intuitive analysis method and a regression analysis method. By visual analysis of the test results, it was found that the minimum value of warpage occurred in test No. 28 and that it was 0.4008 mm. However, the deformation value does not meet the technical requirement of warping below 0.38mm required by products. Therefore, the process parameters obtained by visual analysis do not meet the requirements, and regression analysis needs to be carried out on the test. In the injection molding process of the plastic part, the molding quality of the plastic part is influenced by the combined action of multiple factors, and the nonlinear and multi-coupling condition is shown, so that a regression model of a quadratic complete polynomial is selected:
where y is the optimization target, n is the number of variables, b0、bi、bii、bijAs a model parameter, xi、xjTo design variables, ε is the precision error.
Considering the interaction of the two factors and the influence factor is 7, the regression model contains 1+2n + n (n-1)/2 parameters, i.e. there are 36 unknowns in the above formula, and it is solved by using spss22.0, and the coefficients of the regression model are obtained as shown in table 2:
TABLE 2 regression model coefficients
Wherein x1To injection pressure, x2To maintain the pressure, x3Is the melt temperature, x4Is the mold temperature, x5For cooling time, x6As dwell time, x7Is the injection time.
The results of the analysis of variance of the regression model are shown in table 3:
TABLE 3 analysis of variance
a.R21- (residual sum of squares)/(corrected sum of squares) 0.999.
Coefficient of determinability R2The closer to 1, the better the established equation expresses the relation between the research object and the research object. But R is2Is easily influenced by the number of research factors, so that the coefficient R can be adjusted by calculationa2：
In the formula, n is the number of samples, and k is the number of parameters in the fitting equation.
Calculated adjusted coefficient of determinability
The coefficient is 0.998, the coefficient of the decision and the coefficient of the adjustment decision are both close to 1, which shows that the fitting accuracy of the established regression equation is higher. Solving the model within the selected process parameter range to obtain the process parameter with the minimum plastic part warping amount as follows: injection pressure (x)1)95MPa, pressure holding pressure (x)2)94MPa, melt temperature (x)3)239 ℃ and the mold temperature (x)4) Cooling at 50 ℃ for a period of time (x)5)10s, dwell time (x)6)29s and injection time (x)7)4s, the predicted warpage amount is 0.301 mm. The optimized injection molding process parameters were input into the Moldflow software to obtain a simulated warpage value of 0.3346mm as shown in FIG. 2. The warping value is reduced by 16.5 percent compared with the minimum value of 0.4008mm obtained by a uniform test, the production requirement of the product is met, but the relative error between the predicted value and the simulation value is 10 percent, the relative error is large, the model prediction precision is deficient, and a better model needs to be searched for predicting the molding quality.
(III) constructing a support vector machine model
A Support Vector Machine (SVM) is a machine learning method proposed by Vapnik for classification and regression, and can solve a series of problems such as small samples, nonlinearity, high dimensional numbers, local minimum and the like. The SVM has been proved to be a very powerful and effective regression technique, aiming at the linear classification problem, the SVM takes the solution of the optimal hyperplane of linear classification as a theoretical basis, and the principle of minimizing the structural risk is realized by adjusting the maximum classification interval; aiming at the problem of nonlinear classification, the SVM uses a kernel function to map a low-dimensional input space to a high-dimensional feature space, and linear classification is carried out in the high-dimensional space to obtain an optimal hyperplane.
SVM is used for regression analysis, which is different from classification in that the output of a model is different, in which the output value is a real number, and the output in the classification problem is a specific few values. SVM is classified into linear regression and nonlinear regression for regression analysis. For the case of nonlinear regression, the original data space needs to be mapped into a high-dimensional space through mapping Φ, and the mapping relationship is as follows:
Φ:x→Φ(x)
and then performing approximate linear regression in a high-dimensional space to realize the nonlinear regression of a low-dimensional space. By introducing a kernel function K (x)i,xj)＝Φ(xi)·Φ(xj) Therefore, complex dot product operation in a high-dimensional space is avoided, and the calculation complexity is effectively reduced.
As shown in fig. 3, the SVM construction steps are as follows:
step 1: determining influence factors and quality indexes according to research objects, optimizing the injection molding process of the shell of the power adapter, and selecting injection time (t)injS), cooling time (t)cS), melt temperature (T)meltDEG C), dwell time (t)hS), holding pressure (P)hMPa), die temperature (T)moldIn deg.C) and injection pressure (P)injMPa) is used as an influencing factor, the variation range of each factor is determined, and the warping deformation is used as a quality index.
Step 2: and obtaining a response value of the shell quality index of the power adapter according to the uniform test and the computer simulation value.
And step 3: and aiming at different units and variable ranges of all the influence factors, normalizing the influence factors and the quality index values to eliminate the influence of dimension on analysis, and taking the normalized data as a training sample and a test sample of the SVM model.
And 4, step 4: and selecting a kernel function of the SVM, and optimizing related parameters of the SVM model so as to obtain the optimal model parameter combination.
And 5: training the SVM model by using the training samples, and checking the precision of the SVM model. Mean Square Error (RMSE), maximum relative error (RMAE) and sample decision coefficient (R) may be employed2) To measure the regression performance of the model. The smaller the mean square error and the maximum relative error, the better the sample decision coefficient is, the closer to 1, when R is2And when the precision of the proxy model is more than or equal to 0.9, the precision of the proxy model can be considered to meet the requirement.
Step 6: and if the precision of the SVM model does not meet the requirement, increasing the number of training samples, turning to the step 2, and retraining the SVM model until an ideal proxy model is obtained.
A Genetic Algorithm (GA) is firstly proposed by Holland and is used for simulating a natural selection process for solving and optimizing a nonlinear complex problem, a flow chart of the genetic algorithm for optimizing SVM parameters is shown in FIG. 4, and an SVM model parameter c is obtained to be 51.0091, and g is obtained to be 0.011158.
Taking 7 factors of the process parameters as input values of the SVM model, taking the warpage deformation as an output value of the SVM model, constructing a support vector machine model, and fitting the relationship between the 7 factors and the warpage deformation, wherein the SVM model is shown in FIG. 5.
Firstly, randomizing 60 groups of data, then training the SVM by using 1-50 groups of randomized test data as training data, and using the rest 10 groups of randomized test data as test data for checking whether an SVM model meets requirements or not; the 60 sets of test data were input into the trained SVM model to obtain the predicted output and compared with the expected output, the result is shown in fig. 6, and the predicted relative error is shown in fig. 7. As can be seen from FIG. 7, the SVM fitting effect meets the test requirements, and can be used for predicting the plastic part warpage value.
(III) particle swarm optimization and verification
Particle Swarm Optimization (PSO) is a swarm intelligence optimization algorithm proposed by Kennedy and Eberhart in 1995. The PSO algorithm is derived from the study of bird predation behavior and has been widely applied to most modern scientific and engineering optimization problems.
The PSO algorithm is inspired from the behavior characteristics of the biological population and used to solve the optimization problem, where each particle in the algorithm is a potential solution to the problem and corresponds to a fitness value determined by a fitness function. The particle speed determines the moving direction and distance of the particles, and the particle speed is dynamically adjusted along with the movement of the particle speed per se and other particles, so that the optimization of an individual in a solvable space is realized.
The PSO algorithm first initializes a population of particles in a solution space, each particle representing a potentially optimal solution to the optimization problem, and uses the position, velocity, and fitness value to represent the characteristics of the particle, the fitness value being calculated by a fitness function, which represents the goodness of the particle. The particles move in solution space by tracking individual extrema PbestAnd group extremum GbestUpdating the individual position; individual extremum PbestIs the optimal fitness value obtained by calculating the position of an individual, i.e. a group extreme value GbestRefers to the optimal fitness value searched by all particles in the population. The fitness value is calculated once for each updating of the particle position, and the individual extreme value P is updated by comparing the new particle fitness value with the fitness values of the individual extreme value and the group extreme valuebestAnd group extremum Gbest。
Suppose that in a D-dimensional search space, there is a population X ═ of n particles (X)1,X2,…,Xn) The ith particle represents a vector X of dimension Di＝[x1,x2,…,xd]TWhich represents the position of the ith particle in the D-dimensional search space. Each particle X can be calculated according to the fitness functioniA corresponding fitness value. In the iterative process, the particles update their own velocity and position through individual extrema and population extrema, as follows:
in the formula, omega is an inertia weight; d ═ 1,2, …, D; 1,2, …, n; k is the current iteration number; vidIs the velocity of the particle; c. C1And c2Is an acceleration factor; r is1And r2Is [0,1 ]]A random number in between; position and speed being limited to the interval [ -X ]max,Xmax]And [ -V [ ]max,Vmax]. An extreme value is found for the SVM model by using the PSO algorithm, the trained SVM model is used as a fitness function, and the optimization flow is shown in FIG. 8.
Searching an extreme value for the support vector machine model through a PSO algorithm, and obtaining the optimal process parameters as shown in FIG. 9 according to the optimization result: 4s of injection molding time, 23.3s of cooling time, 239 ℃ of melt temperature, 29s of dwell time, 94MPa of dwell pressure, 59.5 ℃ of mold temperature and 106.6MPa of injection pressure, the minimum value of warpage is obtained by optimizing and is 0.3315mm, the optimized technological parameters are input into Moldflow software for simulation, the simulation value of warpage is obtained and is 0.3323mm, the simulation result is shown in figure 10, the relative error between the predicted value and the simulation value is 0.24 percent and is smaller than the polynomial prediction error, the prediction is obviously improved compared with 0.4008mm obtained by a uniform test before optimization, the reduction is 17.1 percent, and the SVM model prediction is more accurate.
The invention uses the uniform test, effectively reduces the test times when the horizontal number is more, solves the problem of uneven horizontal division caused by different multi-factor change ranges, and leads the distribution of the test points in the change range to be more uniform; by polynomial regression analysis, the relation between the influence factors and the response values is analyzed, the prediction precision is high, the plastic part warping deformation is effectively reduced, and the product production quality is improved; optimizing the model parameters of the support vector machine through a genetic algorithm, solving the problem of SVM model parameter selection, effectively improving the complex relationship between the mapping factors and the response values of the support vector machine and laying a foundation for optimizing process parameters; searching an extreme value of the trained SVM model by using a particle swarm algorithm to obtain an optimal process parameter, inputting the optimal process parameter obtained by optimizing the PSO algorithm into a Moldflow for simulation, wherein the result shows that the quality index of the optimized product is obviously improved, the optimized process parameter is used for a production test, and the result shows that the optimized process parameter effectively improves the quality of the product and shows that the effectiveness of optimizing the support vector machine model parameter by using a genetic algorithm and combining the particle swarm algorithm to search the optimal process parameter is improved.
Claims (8)
1. An injection molding process optimization method based on a support vector machine and a particle swarm algorithm is characterized in that a Genetic Algorithm (GA) is used for optimizing Support Vector Machine (SVM) parameters and a particle swarm algorithm (PSO) is used for optimizing optimal process parameters, and firstly, process parameters are selected as influence factors according to specific production conditions and quality indexes of injection molding products to design a uniform test; secondly, dividing the uniform test data into training data and test data, optimizing parameters of a support vector machine model by using a genetic algorithm and constructing the support vector machine model to obtain a nonlinear mapping relation between injection molding process parameters and quality indexes; and finally, optimizing the process parameters by combining a particle swarm algorithm to obtain the optimal process parameters.
2. The injection molding process optimization method based on the support vector machine and the particle swarm optimization according to claim 1, wherein during a uniform test, after determining each influence factor and an optimization target, a mixed horizontal uniform test table is horizontally divided according to the variation range of each influence factor, after the mixed uniform test table is generated, the magnitude of the warping amount of the injection molded part under different levels of each influence factor is simulated through Moldflow software, then a regression model is established for regression analysis, and the process parameter with the minimum warping amount of the injection molded part is obtained as a verification index.
3. The injection molding process optimization method based on the support vector machine and the particle swarm optimization according to claim 2, wherein the mixing level uniformity test table is as follows:
wherein n is the number of tests, t1，t2，t3Is the horizontal number of columns, q1，q2，q3Respectively, the horizontal number is t1，t2，t3The number of columns of (c).
4. The injection molding process optimization method based on the support vector machine and the particle swarm optimization according to claim 2, wherein a DPS data system is used to generate the mixing uniformity test table.
5. An injection molding process optimization method based on a support vector machine and a particle swarm algorithm according to claim 2, characterized in that the regression model is a quadratic complete polynomial:
where y is the optimization target, n is the number of variables, b0、bi、bii、bijAs a model parameter, xi、xjTo design variables, ε is the precision error.
6. The injection molding process optimization method based on the support vector machine and the particle swarm optimization of claim 5, wherein when the regression model is established for regression analysis, variance analysis is performed after each coefficient of the regression model is obtained, and the coefficient is adjusted by calculation
The verification is carried out in such a way that,
in the formula, n is the number of samples, and k is the number of parameters in the fitting equation;
and solving the regression model within the selected process parameter range to obtain the process parameter with the minimum plastic part warping amount.
7. An injection molding process optimization method based on a support vector machine and a particle swarm algorithm according to claim 1, wherein the support vector machine model is constructed by the following steps:
step 1: determining various influencing factors and quality indexes according to the material and the actual production condition of the plastic part, and determining the variation range of the various influencing factors;
step 2: obtaining a response value of the quality index of the plastic part according to a uniform test and a computer simulation value;
and step 3: aiming at different units and variable ranges of all the influence factors, carrying out normalization processing on the influence factors and the quality index values to eliminate the influence of dimensions on analysis, and taking normalized data as a training sample and a test sample of a support vector machine model;
and 4, step 4: selecting a kernel function of the support vector machine model, and optimizing related parameters of the support vector machine model to obtain the optimal model parameter combination;
and 5: training a support vector machine model by using a training sample, and checking the precision of the support vector machine model;
step 6: and if the accuracy of the support vector machine model does not meet the requirement, increasing the number of training samples, turning to the step 2, and retraining the support vector machine model until an ideal support vector machine model is obtained.
8. The injection molding process optimization method based on the support vector machine and the particle swarm optimization according to claim 1, wherein the particle swarm optimization is used for searching an extreme value of the support vector machine model, and the trained support vector machine model is used as a fitness function to be optimized to obtain the optimal process parameters.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910789733.0A CN110674598A (en) | 2019-08-26 | 2019-08-26 | Injection molding process optimization method based on support vector machine and particle swarm optimization |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910789733.0A CN110674598A (en) | 2019-08-26 | 2019-08-26 | Injection molding process optimization method based on support vector machine and particle swarm optimization |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110674598A true CN110674598A (en) | 2020-01-10 |
Family
ID=69075769
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910789733.0A Pending CN110674598A (en) | 2019-08-26 | 2019-08-26 | Injection molding process optimization method based on support vector machine and particle swarm optimization |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110674598A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111597729A (en) * | 2020-05-27 | 2020-08-28 | 北京天泽智云科技有限公司 | Method and system for optimizing technological parameters of processing equipment |
| CN112508220A (en) * | 2020-09-22 | 2021-03-16 | 内蒙古工业大学 | Traffic flow prediction method and device |
| CN112906155A (en) * | 2021-02-05 | 2021-06-04 | 深圳市浦联智能科技有限公司 | Virtual measurement method for injection molding product information |
| CN113119425A (en) * | 2021-03-22 | 2021-07-16 | 广东工业大学 | Injection molding product quality prediction method based on improved support vector machine |
| CN113427730A (en) * | 2020-03-23 | 2021-09-24 | 住友重机械工业株式会社 | Injection molding machine and state determining device |
| CN113591339A (en) * | 2021-06-23 | 2021-11-02 | 江苏师范大学 | Modeling method capable of predicting temperature rise and thermal error of ball screw |
| CN115965166A (en) * | 2023-03-16 | 2023-04-14 | 昆山市恒达精密机械工业有限公司 | Optimization method and system for plastic product production process |
| CN116777537A (en) * | 2023-05-26 | 2023-09-19 | 华能国际电力股份有限公司上海石洞口第二电厂 | Method and system for calculating and optimizing electricity-measuring coal cost and blending coal based on coal quality characteristics |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110073301A (en) * | 2017-08-02 | 2019-07-30 | 强力物联网投资组合2016有限公司 | The detection method and system under data collection environment in industrial Internet of Things with large data sets |
| CN110367994A (en) * | 2019-08-16 | 2019-10-25 | 福建工程学院 | Piezoelectric matrix actively monitoring gait device and its working method |
-
2019
- 2019-08-26 CN CN201910789733.0A patent/CN110674598A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110073301A (en) * | 2017-08-02 | 2019-07-30 | 强力物联网投资组合2016有限公司 | The detection method and system under data collection environment in industrial Internet of Things with large data sets |
| CN110367994A (en) * | 2019-08-16 | 2019-10-25 | 福建工程学院 | Piezoelectric matrix actively monitoring gait device and its working method |
Non-Patent Citations (2)
| Title |
|---|
| 赵斐: "注塑过程质量控制中若干关键问题的研究", 《中国优秀硕士学位论文全文数据库》 * |
| 黄海跃等: "基于神经网络和遗传算法的薄壳塑件注塑工艺优化", 《塑料》 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113427730A (en) * | 2020-03-23 | 2021-09-24 | 住友重机械工业株式会社 | Injection molding machine and state determining device |
| CN113427730B (en) * | 2020-03-23 | 2023-12-05 | 住友重机械工业株式会社 | Injection molding machine and state determination device |
| CN111597729A (en) * | 2020-05-27 | 2020-08-28 | 北京天泽智云科技有限公司 | Method and system for optimizing technological parameters of processing equipment |
| CN111597729B (en) * | 2020-05-27 | 2023-07-25 | 北京天泽智云科技有限公司 | Processing equipment technological parameter optimization method and system |
| CN112508220A (en) * | 2020-09-22 | 2021-03-16 | 内蒙古工业大学 | Traffic flow prediction method and device |
| CN112906155A (en) * | 2021-02-05 | 2021-06-04 | 深圳市浦联智能科技有限公司 | Virtual measurement method for injection molding product information |
| CN113119425A (en) * | 2021-03-22 | 2021-07-16 | 广东工业大学 | Injection molding product quality prediction method based on improved support vector machine |
| CN113591339A (en) * | 2021-06-23 | 2021-11-02 | 江苏师范大学 | Modeling method capable of predicting temperature rise and thermal error of ball screw |
| CN113591339B (en) * | 2021-06-23 | 2024-03-22 | 广东智目科技有限公司 | Modeling method capable of predicting temperature rise and thermal error of ball screw |
| CN115965166A (en) * | 2023-03-16 | 2023-04-14 | 昆山市恒达精密机械工业有限公司 | Optimization method and system for plastic product production process |
| CN116777537A (en) * | 2023-05-26 | 2023-09-19 | 华能国际电力股份有限公司上海石洞口第二电厂 | Method and system for calculating and optimizing electricity-measuring coal cost and blending coal based on coal quality characteristics |
| CN116777537B (en) * | 2023-05-26 | 2024-04-12 | 华能国际电力股份有限公司上海石洞口第二电厂 | Method and system for calculating and optimizing electricity-measuring coal cost and blending coal based on coal quality characteristics |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110674598A (en) | Injection molding process optimization method based on support vector machine and particle swarm optimization | |
| Fei et al. | Springback prediction for incremental sheet forming based on FEM-PSONN technology | |
| CN110640982A (en) | Multi-target optimization method for injection molding process parameters of thin-wall injection molding part | |
| CN111563347B (en) | Injection molding process parameter optimization method for fiber reinforced composite material | |
| CN112115579A (en) | Multi-target optimization method for injection molding process parameters of glass fiber reinforced plastics | |
| Cao et al. | Multi-objective optimization of injection-molded plastic parts using entropy weight, random forest, and genetic algorithm methods | |
| CN112101630A (en) | Multi-target optimization method for injection molding process parameters of thin-wall plastic part | |
| CN112069610B (en) | Injection molding process parameter optimization method for transparent complex multi-cavity plastic part | |
| CN106296434B (en) | Grain yield prediction method based on PSO-LSSVM algorithm | |
| CN113191050B (en) | Workpiece deformation simulation prediction method based on MSVR-GA (minimum shift register-GA) optimized initial residual stress | |
| CN111310348A (en) | Material constitutive model prediction method based on PSO-LSSVM | |
| CN113103535A (en) | GA-ELM-GA-based injection molding part mold parameter optimization method | |
| CN113158368A (en) | Prediction and control method and device for deformation of material-increasing and material-decreasing composite manufacturing part | |
| CN111286599A (en) | Alloy heat treatment process optimization method based on machine learning | |
| Liu et al. | Multiobjective optimization of injection molding process parameters for the precision manufacturing of plastic optical lens | |
| CN114274457A (en) | Kriging-GA-based structural parameter decision method for injection mold pouring system | |
| CN111369039B (en) | Alloy heat treatment process optimization method based on multi-objective optimization assisted active learning | |
| CN111444619B (en) | Online analysis method and equipment for injection mold cooling system | |
| CN112580855A (en) | Cable group steady-state temperature rise prediction method based on self-adaptive variation PSO-BP neural network | |
| Gao et al. | A 6-sigma robust optimization method for stamping forming of automobile covering parts based on residual error and radial basis interpolation | |
| Luo et al. | Iterative improvement of parameter estimation for model migration by means of sequential experiments | |
| CN116894379A (en) | Injection molding quality optimization method | |
| CN114297582A (en) | Modeling method of discrete counting data based on multi-probe locality sensitive Hash negative binomial regression model | |
| Chen et al. | Multi-objective optimal approach for injection molding based on surrogate model and particle swarm optimization algorithm | |
| CN114692507A (en) | Counting data soft measurement modeling method based on stacking Poisson self-encoder network |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200110 |