The Injection Molding Process Parameters Optimization Of Radiator Shell Based On Support Vector Machine And Particle Swarm Optimization

Injection molded products are widely used in life,machinery and electronics,and with the continuous development of the times,people have higher and higher requirements for the quality of injection molded products.Because the quality of injection molding is related to many factors,it has the characteristics of multiple parameters and strong nonlinearity.The relationship between injection molding quality and molding factors is complex.Therefore,the relationship between influencing factors and quality indicators is a typical black box problem.When optimizing the process parameters,the test points cannot be evenly distributed due to the different ranges of each process parameter,and the spans between levels are greatly different.In addition,due to the limited number of trials,the method based on the empirical risk minimization theory cannot meet the requirements for an infinite amount of data.Therefore,this article uses uniform experimental design for experimental design to ensure that the levels are evenly distributed in the range of change,and a support vector machine(SVM)based on the theory of structural risk minimization is used to map the complex relationship between influencing factors and various quality indicators.Based on the theory of structural risk minimization,the SVM can process the data of a limited sample,effectively solve the above problems,and particle swarm optimization(PSO)is used to optimize the injection molding process parameters.The radiator shell is taken as the research object in the paper,and the injection molding process parameters are optimized to solve the problems of warpage and shrinkage during the injection molding process.First,based on Moldflow finite element analysis software,a finite element analysis model of the radiator shell is established and simulated.Secondly,experiments are designed based on the uniform experimental design method,and the obtained different levels of test parameters are input into Moldflow finite element analysis software for simulation analysis to obtain warpage deformation and volume shrinkage rate under different process parameter combinations.Then nonlinear regression analysis was performed on the training data based on SPSS software to obtain quadratic polynomials for warpage deformation and volume shrinkage rate,respectively.The accuracy of the quadratic polynomial regression model is tested with test data.When using SPSS software to perform non-linear regression analysis on training data,although the determination coefficient of the training data is good,the test data is input to the quadratic polynomial regression model during the test,the relative error of some data was more than 5%.It can be seen that the generalization ability of the quadratic polynomial regression model is insufficient.Therefore,based on the data of uniform experimental design,the influence of the dimension on the regression is removed through the normalization of the test data,and the genetic algorithm(GA)is used to optimize the model parameters of the SVM,and the best model parameters of the SVM are obtained.Thus,the SVM models of the warpage deformation and volume shrinkage rate of the plastic parts of the radiator shell are established respectively.The relative errors of the SVM model established by the test are below 5% in the training samples and the test samples.It has been greatly improved compared with the quadratic polynomial regression model,which can more accurately reflect the relationship between influencing factors and warpage deformation and volume shrinkage rate.Finally,the SVM models of warpage deformation and volume shrinkage rate are searched iteratively by the PSO to achieve the optimized injection molding process parameters for different quality problems.For the quality index of warpage deformation,when the process parameters are melt temperature of 239?,mold temperature of 72.5?,injection pressure of 112.7MPa,injection time of 4s,cooling time of 29 s,holding pressure of 94 MPa,and holding time of 29 s,the predicted warpage deformation value is 0.3337 mm.The process parameters obtained through optimization are input to Moldflow software for simulation.The warped simulation value is 0.3307 mm,and the relative error between the warped simulation value and the predicted value is 0.91%.For the quality index of volume shrinkage rate,when the process parameters are melt temperature of 210?,mold temperature of 79?,injection pressure of 154 MPa,injection time of 2.4s,cooling time of 10 s,holding pressure of94 MPa,and holding pressure of 29 s,the predicted volume shrinkage rate is 2.5028%,and the simulation value of the volume shrinkage rate is 2.596%,and the relative error between the simulation value of the volume shrinkage rate and the predicted value is3.59%.It shows that the generalization ability of the constructed SVM model is good and it can accurately reflect the relationship between process parameters and quality indicators.Considering the two quality indicators of warpage deformation and volume shrinkage rate,and comparing the warpage and volume shrinkage rate under the optimized process parameters,it is determined that the SVM model has the best results for the process parameters obtained by the optimization of warpage.Finally,it is verified through experiments that the optimization of the injection molding process parameters based on SVM and PSO is reliable and effective.