Optimization and control of filter element bracket injection molding quality based on RBF and LSGRG
Aiming at the problems of large volume shrinkage rate and high production energy consumption of filter element bracket of complex staggered thin wall grille for automobile,finite element simulation method was used to establish the finite element model of filter element bracket simulation analysis of complex staggered thin wall grille,the volume shrinkage rate of plastic parts and the maximum clamping force in the injection process were optimized,so as to improve the molding quality of plastic parts and reduce the production energy consumption.Firstly,a prediction model was established based on optimal latin hyper-cube sampling(OLHS)method and radial basis function(RBF)neural network model,and the accuracy of the model was verified.Secondly,by analyzing 20 groups of calculation data and combining with engineering experience,it was found that the volume shrinkage rate and the maximum clamping force are in conflict with each other,and have a great impact on molding quality and energy consumption.For this reason,large scale generalized reduced gradient(LSGRG)technology was adopted.LSGRG was used for multi-objective optimization of the prediction model and the optimal combination of process parameters was obtained.The maximum mode-locking force optimized by LSGRG method was reduced by 10.30%,and the volume shrinkage rate was reduced by 27.61%.The optimization effect was significant.Finally,the filter element bracket with complexstaggered thin wall grating was produced by using the optimized injection molding process parameters on the injection molding machine,and the product quality met the requirements.The joint multi-objective optimization method based on RBF neural network model and LSGRG technology proposed in this paper can provide reference for the molding quality control and optimization of the same type of plastic parts.
filter element bracketmulti-objective optimizationoptimal latin hypercube samplingradial basis function neural networkLSGRG technique
NETL
NSTL
万方数据
