基于CAE和BP神经网络的塑件气孔及注塑工艺优化分析

摘要

针对除霜格栅塑件注塑后由于气孔较多而引起的开裂、凹陷等问题,对气孔产生的成因进行了分析.在排除材料、模具影响因素后,针对注塑成型工艺因素中的注塑速度和注塑压力对气孔问题产生的影响,在基于CAE仿真分析的基础上,将注塑速度和注塑压力转化成相应的螺杆转速控制因素,结合正交试验法对控制因素进行分层,通过BP神经网络构建控制因素与气孔数量的非线性控制关系,通过BP神经网络的预测作用,寻优出气孔最少的控制因素水平组合,并将之反馈于CAE仿真进行验证计算,检验结果表明,所寻优出的工艺参数水平组合能将气孔数量控制在较低的数量上.通过上述寻优,找到了改善塑件气孔的注塑工艺方案,对BP神经网络应用于注塑成型的优化具有很好的参考价值.%Aiming at the problems of cracking and depression caused by the porosity of plastic injection part defrost grid,the causes of the formation of the pores were analyzed.After eliminating the influence factors of material and mold,the influence of injection molding process of injection speed and injection pressure on the porosity,the injection speed and injection pressure convert into the corresponding control factor of screw speed based on the computer aided engineering (CAE) simulation analysis.Then the orthogonal test was stratified on control factors,the nonlinear relationship between control factors and the number of holes was constructed through back propagation (BP) neural network.And through BP neural network prediction,the optimal combination of control factor level for the least hole number was found,and the feedback from the CAE simulation verified the calculation.The testresults show that the optimized process parameters combination can control the number of stomata.Through the above optimization,the injection molding process to improve the product porosity was found.The application of BP neural network in the optimization of injection molding has a good reference value.