基于BP神经网络的汽车内饰面板注塑成型工艺参数优化

Optimization of Injection Molding Process for Automotive Interior Panel Based on BP Neural Network

以汽车内饰件中立柱上面板注塑成型为例,建立了模流CAE分析模型,运用Moldflow 2015软件对注塑成型工艺参数进行仿真,对注塑过程中的翘曲原因进行了分析;结合塑件的翘曲优化目标,提出了一种结合Tugachi正交试验法、BP神经网络、Matlab数值分析改善产品翘曲变形的注塑成型工艺参数寻优方法,基于此方法对注塑成型工艺参数进行了多次优化,并对优化结果进行了CAE模流分析验证。结果表明:神经网络预测结果与CAE模流分析结果相近,塑件最小翘曲量能降低至1.497 mm,对应的注塑成型工艺参数为:T_θ(205℃)、T_s(40℃)、P_I(60 MPa)、t_i(2.2 s)、P_(h1)(85 MPa)、t_(h1)(11.5 s)、P_(h2)(30 MPa)、t_(h2)(7 s)、t_c(20 s),将最终寻优所得参数输入注塑机,经试模验证后,产品注塑翘曲得到改善,与CAE分析预期值接近;提出的注塑参数优化设计方法能有效降低模具试模成本,缩短模具生产周期。

The injecting molding craftwork for upright panel part of the automotive interior parts was given. The Moldflow CAE analysis model was established, the injection molding process parameters were simulated by Moldflow 2015 software. The reason of the warp in the process of injection molding was analyzed. Combined with the optimization target of the plastic part , an optimization method of injection molding process parameters by using Tugachi orthogonal test method, BP neural network and Matlab numerical analysis was presented. Based on this method, the injection molding process parameters were optimized, and the optimization results were verified by CAE Moldflow analysis. The results showed that the prediction results of neural network were similar to that of CAE Moldflow analysis. The plastic part had been reduced to 1. 497 mm, and the better molding process parameters of plastic parts as follow was To （ 205 ℃）, T （40 ℃ ）, P1 （ 60 MPa）, ti （ 2.2 s）, Ph, （ 85 MPa）, th1 （ 11.5 s） ,Ph2 （ 30 MPa） , th2 （7 s） , tc （ 20 s） ,The resulting parameters of input to the injection molding machine, after the test moding validation, the warp was improved, and CAE analysis was expected to be close to the expected value. The proposed optimization design method could effectively reduce the cost of mould and shorten the production cycle of the mould.