摘要
电阻点焊是一个多变量相互作用的复杂物理过程,点焊质量容易受很大的不确定性影响,因此本文提出了结合WOA优化算法和XGBoost算法进行电阻点焊熔核直径预测。该模型将主要的工艺参数和板材厚度作为模型的特征输入,将超声波检测的焊点熔核直径作为输出,构建了一个包含102个由7个输入属性组成的实验实例的数据库,用于训练和测试XGBoost回归模型;利用WOA优化算法寻找XGBoost的最优结构,并对WOA优化算法进行改进以提高模型预测准确性。结果表明,WOA-XGBoost模型相较于其他单一机器学习模型具有更高的预测精度。该组合模型可以帮助企业分析影响熔核直径大小的特征因素,同时为调节工艺参数提供理论依据,有望在汽车焊接过程中发挥重要作用。
Resistance spot welding is a complex physical process characterized by multifactorial interactions,leading to inherent uncertainty in the quality of spot welds.Therefor,this study proposed a novel approach that combined the whale optimization algorithm(WOA)with the XGBoost algorithm for predicting the nugget diameter in resistance spot welding.This predictive model leveraged crucial process parameters and material thickness as input features,while utilizing ultrasonic inspection-measured nugget diameter as the output variable.A database consisting of 102 experimental instances,each composed of seven input attributes,was constructed for training and testing the XGBoost regression model.Furthermore,the WOA optimization algorithm was employed to identify the optimal structure of the XGBoost model,and improved the WOA optimization algorithm to enhance the model prediction accuracy.The results show that the WOA-XGBoost model outperforms alternative standalone machine learning models,demonstrating higher prediction precision on the test set.This integrated model holds promise in assisting industries to scrutinize the influential factors affecting nugget diameter and offering a theoretical foundation for process parameter adjustments,thereby,potentially playing a pivotal role in automotive welding procedures.
作者
郑银环
宋宣科
卢杰
丁刚强
ZHENG Yinhuan;SONG Xuanke;LU Jie;DING Gangqiang(School of Mechanical and Electrical Engineering,Wuhan University of Technology,Wuhan 430070,China;Liuzhou Wuling New Energy Vehicle Co.,Ltd.,Liuzhou 545007,China)
出处
《热加工工艺》
北大核心
2024年第24期53-60,共8页
Hot Working Technology
基金
基于5G网络和云平台的新能源汽车智能制造关键技术集成与应用项目(桂科AA21077016)
柳州科技计划重点项目(2021AA0101A019)。
关键词
电阻点焊
集成学习
WOA优化算法
熔核直径
resistance spot welding
ensemble learning
whale optimization algorithm(WOA)
nugget diameter
