摘要
利用半监督机器学习模型,对铝合金的成分-屈服强度、成分-抗腐蚀性数据进行扩充。完成数据扩充后,使用得到的完整成分-屈服强度-抗腐蚀性数据进行有监督的机器学习模型训练。该机器学习模型既可根据铝合金成分预测其性能,也可以根据目标屈服强度和抗腐蚀性反向设计铝合金的成分。结果表明:模型依据铝合金成分预测的性能数据与真实数据相差不超过10%;且模型根据两种目标性能设计出的铝合金成分与真实铝合金成分非常相近,未来可以使用该方法设计具有高屈服强度与抗腐蚀性的新型铝合金。
The data of composition-yield strength and composition-corrosion resistance of aluminum alloy are expanded by using semi-supervised machine learning model.After the data expansion is completed,supervised machine learning model training is performed by using the complete composition-yield strength-corrosion resistance data obtained.The machine learning model can be used not only to predict the properties of the aluminum alloy based on its composition but also to reversely design the composition of the aluminum alloy based on the target yield strength and corrosion resistance.The results show that the difference between the performance data predicted by the model based on aluminum alloy composition and the real data is less than 10%.The aluminum alloy composition designed by the model according to the two target properties is very close to the real aluminum alloy composition,and this method can be used to design a new aluminum alloy with high yield strength and corrosion resistance in the future.
作者
定巍
苗以升
张坛
黄丽颖
Ding Wei;Miao Yisheng;Zhang Tan;Huang Liying(Inner Mongolia University of Science and Technology,Baotou,Inner 014010,China;Hebei Vocational University of Technology and Engineering,Xingtai,Hebei 054035,China)
出处
《邢台职业技术学院学报》
2023年第5期46-53,共8页
Journal of Xingtai Polytechnic College
基金
内蒙古科技大学教学改革研究项目--“大数据时代下工科大学生科学与工程素质训练计划(SSET)-从科学研究与工程实践的数据收集出发”,项目编号:JY2021026
内蒙古自然科学基金--“基于薄膜奥氏体控制和提高贝氏体相变速度的热镀锌TRIP钢原始组织设计”,项目编号2020LH05026
河北省高等学校科学技术研究项目--“非标装备用碳钢搅拌摩擦焊接关键技术研究”,项目编号:ZD2022034
邢台市青年人才计划项目--“搅拌摩擦加工可控降解镁合金组织调控及体外腐蚀机理研究”项目编号:2021ZZ033
邢台市科技人才专项项目--“SS304和Q235异质钢搅拌摩擦焊接基础研究”,项目编号:2022zz115。
关键词
半监督机器学习
铝合金
屈服强度
抗腐蚀性
semi-supervised machine learning
aluminum alloy
yield strength
corrosion resistance
