论文部分内容阅读
提出了一种改进多目标粒子群优化(IMOPSO)算法,并用于优化注塑成型过程中熔接痕的长度和相遇角。基于成型工艺参数建立了熔接痕多目标优化模型,同时提出了改进混合神经网络(HNN)作为预测熔接痕长度和相遇角的代理模型。其中,通过Taguchi方法设计实验,采用Moldflow软件得到了训练改进HNN的样本。基于Pareto支配理论,提出了一种IMOPSO算法,并通过算例验证了其在多目标优化问题中的有效性。采用IMOPSO算法对注塑件熔接痕的长度和相遇角进行优化。将优化结果和MPI实验结果进行比较表明,IMOPSO算法能有效地优化注塑制品的熔接痕质量。
An improved multi-objective particle swarm optimization (IMOPSO) algorithm was proposed and used to optimize the length and angle of weld penetration during injection molding. Based on the forming process parameters, a multi-objective optimization model of weld line is established. At the same time, an improved hybrid neural network (HNN) is proposed as a proxy model for predicting weld line length and meeting angle. Among them, through the Taguchi method design experiments, using the Moldflow software to get training to improve the HNN samples. Based on Pareto domination theory, an IMOPSO algorithm is proposed, and its validity in multi-objective optimization problems is verified by an example. The IMOPSO algorithm is used to optimize the length and angle of fusion of the weldments. The comparison between the optimized results and the MPI results shows that the IMOPSO algorithm can effectively optimize the weld line quality of injection molded products.