基于机器学习的轻合金压铸关键工艺参数分析
Key Parameter Analysis of Die-casting Light Alloy Based on Machine Learning
王鑫 1汪星辰 1彭立明 1付彭怀1
作者信息
- 1. 上海交通大学材料科学与工程学院,轻合金精密成型国家工程研究中心,上海 200240
- 折叠
摘要
通过对新能源汽车前机舱压铸生产线数据的采集,建立了"工艺参数-下线质量"大数据集,并利用4种机器学习模型对大数据集进行训练.结果表明,袋装树模型的预测准确率与泛化效果最好,获得了容忍准确率为77.3%的测试集预测效果.此外,通过计算15种关键工艺参数的相对重要度与灵敏度,获得了工艺参数对铸件性能影响程度权重排序,对压铸工艺参数优化与控制具有积极意义.
Abstract
"Parameters-Offline quality"big database was established by collecting the data from the die-casting produc-tion line of forward engine room of new energy vehicles.Then,four types of machine learning models were used to train the dataset.The results indicate that the bagged decision trees model has the satisfied prediction accuracy and gen-eralization ability,and the tolerated accuracy of prediction results of the test dataset reaches 77.3%.Furthermore,the influences of key parameters on the quality of castings were ranked by calculating the relative influences and sensitivity levels,which has important guiding significance in optimizing and controlling of pressure die-casting parameters.
关键词
机器学习/数据驱动/袋装树模型/关键工艺参数/轻合金Key words
Machine Learning/Data Driven/Bagged Decision Trees Model/Key Parameters/Light Alloys引用本文复制引用
基金项目
国家重点研发计划资助项目(2022YFB3706800)
出版年
2024
NETL
NSTL
万方数据