基于不同机器学习算法的铝合金性能预测  

Prediction of Aluminum Alloy Properties Based on Different Machine Learning Algorithms

作　　者：李婷 LI Ting

机构地区：[1]沈阳东创贵金属材料有限公司,辽宁沈阳110000

出　　处：《有色设备》2023年第4期66-71,82,共7页Nonferrous Metallurgical Equipment

摘　　要：铝合金由于其高强度、低重量的特点被广泛的应用于航空航天以及交通等领域,高性能铝合金的设计是当下的热点。本文以铝合金组分为输入向量,抗拉强度为目标变量,建立了RF、ET、Bagging、Adaboost四种不同的机器学习算法模型。结果表明:RF模型具有最佳的预测性能,R=0.89、MAE=40.33;Ti元素含量对铝合金抗拉强度的预测起正向作用,Ti元素含量越高,抗拉强度值越大;Mg元素、Cu元素含量铝合金抗拉强度的预测作用并不明显;Zn元素、Ce元素、Y元素含量对铝合金抗拉强度的预测起负向作用,即元素含量越大,抗拉强度值越小,特征重要性从大到小分别为Ti>Mg>Cu>Zn>Ce>Y。Aluminum alloys are widely used in aerospace and transportation fields due to their high strength and low weight.In this paper,with aluminum alloy components as the input vector and tensile strength as the target variable,four different machine learning algorithm models,RF,ET,Bagging and Adaboost,are established.The results show that RF model has the best prediction performance,R=0.89,MAE=40.33.The content of Ti plays a positive role in predicting the tensile strength of aluminum alloys.The higher the content of Ti,the greater the tensile strength.The prediction of tensile strength of aluminum alloy with Mg and Cu content is not obvious.The content of Zn element,Ce element and Y el-ement plays a negative role in predicting the tensile strength of aluminum alloy.The element content is great,that the tensile strength value is small.The importance of the characteristics is Ti>Mg>Cu>Zn>Ce>Y.

关 键 词：机器学习 铝合金 抗拉强度 性能预测 

分 类 号：TF821[冶金工程—有色金属冶金]