中国塑料 ›› 2015, Vol. 29 ›› Issue (09): 54-59 .DOI: 10.19491/j.issn.1001-9278.2015.09.011

• 加工与应用 • 上一篇    下一篇

基于EBF神经网络和粒子群算法的注射成型优化设计

张俊红1,陈孔武1,王健2,郭迁2,马梁2   

  1. 1. 天津市南开区天津大学2. 天津大学
  • 收稿日期:2015-03-23 修回日期:2015-06-15 出版日期:2015-09-26 发布日期:2015-09-26

Optimization Design of Injection Molding Based on EBF Neural Network and Particle Swarm Algorithm

  • Received:2015-03-23 Revised:2015-06-15 Online:2015-09-26 Published:2015-09-26

摘要: 基于拉丁超立方设计建立了椭球基(EBF)神经网络模型描述注塑工艺参数与翘曲值间的函数关系,将EBF神经网络模型与Kriging模型对比,说明EBF神经网络模型可以准确地描述注塑工艺参数与翘曲值之间的函数关系,并结合多目标粒子群算法对工艺参数进行优化,并与邻域培植遗传算法优化结果对比,说明多目标粒子群算法的优点。结果表明,基于EBF神经网络模型和粒子群优化算法可以使塑料出水管翘曲值减小11.64 %,同时使保压时间和冷却时间总和减小了2.13 s,从而在出水管批量生产过程中减少了生产时间。

关键词: 翘曲分析, 神经网络, 粒子群算法, 优化

Abstract: The EBF neural network model was established to describe the relationship between the process parameters and warpage based on the Latin hypercube method. Compared with Kriging model, the EBF neural network model could more accurately describe the relationship between the process parameters and warpage. The process parameters were optimized with the EBF model combined with multiobjective particle swarm algorithm, and the result was compared with Neighborhood Cultivation Genetic Algorithm. It was indicated that the injection parameter optimization method based on the EBF neural network and multiobjective particle swarm algorithm approach was feasible, and the warpage was decreased by 11.64 %,production time was shortened because the packing time and cooling time were decreased by 2.53 s.

Key words: warpage analysis, neural network, particle swarm algorithm, optimization