基于PSO-BP神经网络优化的汽车斗框注塑成型优化

Injection Molding Optimization of Car Bucket Frame Based on PSO-BP Neural Network

针对塑件在注塑成型时,翘曲变形量和收缩率均较大的实际生产难题,利用神经网络预测的方法,对其注塑工艺参数进行优化,获得优质的塑件成品。采用粒子群算法(PSO)对BP神经网络进行了改进,并基于神经网络构建了注塑工艺参数与翘曲变形、体积收缩之间的预测模型。在准确地预测了翘曲变形和收缩率的最小变量的基础上,获得了最佳注塑成型工艺参数为(t1、T2、T3、p4、t5、t6)=(3.345 s、71.85℃、213.36 MPa、68%、12.5 s、13.7 s),相应的质量指标(Δq、Δν)的灰色关联适用度为0.089 5,相应的指标为[1.125,3.54%]。翘曲变形能控制在0.95 mm以下,收缩率降低至4%以下,工艺参数优化后的塑件的成型质量同样得到有效控制,大幅缩短了模具的生产周期,提高了模具生产经济效益。

Aimed at the practical production problem of large warpage deformation and shrinkage of plastic part in injection molding,it was necessary to optimize the injection process parameters by using neural network prediction method to obtain high quality plastic part. Firstly,the BP neural network was improved by using particle swarm optimization( PSO),and the prediction model based on neural network between injection molding process parameters and warpage deformation and volume shrinkage was constructed. On the basis of accurately predicting the minimum variable results of warpage deformation and shrinkage,the optimum injection molding process parameters were obtained. The process parameters were( t1,T2,T3,p4,t5,t6) =( 3. 345 s,71. 85 ℃,213. 36 MPa,68%,12. 5 s,13. 7 s). The grey relational applicability of corresponding quality indexes( Δq,Δν) was 0. 089 5. The corresponding index was [1. 125,3. 54%]. Warpage deformation can be effectively controlled below 0. 95 mm and shrinkage could be effectively reduced to less than 4%. The forming quality of the plastic part optimized by the process parameters was also effectively controlled,which greatly shortens the manufacturing cycle of the mould and improves the economic benefits of the mould production.