Multi-objective optimization of injection molded parts with insert based on IFOA-GRNN-NSGA-II
Chunxiao Li,Xiying Fan,Yonghuan Guo,Xin Liu,Changjing Wang,Dezhao Wang
DOI: https://doi.org/10.1515/polyeng-2021-0242
IF: 1.624
2022-04-12
Journal of Polymer Engineering
Abstract:Abstract The physical properties of plastic products, such as local strength, wear resistance and electrical properties, can be improved by adding embedded parts in the appropriate position of the products, and the precision of plastic parts can also be improved. However, due to the addition of inserts, the flow and shrinkage around inserts will be affected. Compared with traditional injection molding products, the quality is difficult to predict. To solve this problem, the injection molded parts with inserts (electrostatic test box) was used as an example, according to the product structure, three objectives of volume shrinkage, warpage in the X direction, and warpage in the Z direction were optimized. A generalized regression neural network (GRNN) model was established with molding parameters as input and quality objectives as output. Improved fruit fly optimization algorithm (IFOA) was proposed to select the optimal smoothing parameters dynamically. Through the prediction of samples, the experimental results show that the model is superior to two comparative models. Non-dominated sorting genetic algorithm (NSGA-II) was used to solve the model, and the Pareto-optimal front was obtained. The entropy TOPSIS method was used to evaluate the Pareto-optimal front, and the optimal solution was obtained. The results show that IFOA-GRNN-NSGA is a reliable multi-objective optimization method.
polymer science