基于机器学习与遗传算法的注塑产品工艺参数优化  

Process Parameter Optimization of Injection Molding Product Based on Machine Learning and Genetic Algorithm

作　　者：龚川 项薇[1,2,3] 陈昱 Gong Chuan;Xiang We;Chen Yu

机构地区：[1]宁波大学机械工程与力学学院,浙江宁波315211 [2]浙江省零件轧制成形技术研究重点实验室,浙江宁波315211 [3]宁波大学先进储能技术与装备研究院,浙江宁波315211

出　　处：《机械制造》2024年第5期60-65,共6页Machinery

基　　金：宁波市自然科学基金资助项目(编号:202003N4154)。

摘　　要：注塑成型工艺可以高效成型高精度产品,被广泛使用。通过优化注塑产品工艺参数,可以获得更高的表面质量和尺寸精度。为了获得最佳工艺参数组合,设计了基于关键工艺参数组合的极端梯度提升决策树模型。为了解决训练数据不平衡问题,引入尺寸加权修正方法,提高对不合格产品尺寸的预测精度。使用遗传算法对改进极端梯度提升决策树模型以产品尺寸误差最小为目标进行全局寻优,得到最佳工艺参数组合。最佳工艺参数组合为模腔压力39.4 bar、切换压力33.8 bar、射嘴压力49.4 bar、回水温度72.6℃、模温机水流量58.7 cm^(3)/s。The injection molding process can efficiently mold high precision product and is widely used.By optimizing the process parameter of injection molded product,higher surface quality and dimensional accuracy can be obtained.In order to obtain the optimal combination of process parameters,the xgboost model based on the combination of key process parameters was designed.In order to solve the problem of unbalanced training data,a size weighted correction method was introduced to improve the prediction accuracy of the size of unqualified product.The genetic algorithm was used to optimize the improved xgboost model with the goal of minimizing the product size error,and the optimal combination of process parameters was obtained.The optimal combination of process parameters is cavity pressure of 39.4 bar,switching pressure of 33.8 bar,nozzle pressure of 49.4 bar,return water temperature of 72.6℃,and water flow rate of 58.7 cm^(3)/s.

关 键 词：机器学习 遗传算法 注塑 参数 优化 

分 类 号：TH162[机械工程—机械制造及自动化]