基于机器学习的Cu-Ni-Co-Si合金固溶处理晶粒尺寸预测  

Prediction of Grain Size in Cu-Ni-Co-Si Alloy Solid Solution Treatment Based on Machine Learning

作　　者：闫碧霄 朱雪彤 吴钢 陈慧琴[1] YAN Bixiao;ZHU Xuetong;WU Gang;CHEN Huiqin(School of Materials Science and Engineering,Taiyuan University of Science and Technology,Taiyuan 030024,China)

机构地区：[1]太原科技大学材料科学与工程学院,山西太原030024

出　　处：《铜业工程》2024年第2期18-22,共5页Copper Engineering

基　　金：中央引导地方科技发展资金自由探索类项目(YDZJSX2021A039);山西省研究生优秀创新项目(2022Y667)资助。

摘　　要：Cu-Ni-Co-Si合金在固溶处理后的晶粒尺寸会影响其服役性能。采用3种不同的机器学习方法——BP神经网络、随机森林、长短期记忆网络,分别建立了Cu-Ni-Co-Si合金固溶温度和固溶时间对固溶后晶粒尺寸影响的机器学习预测模型。对比分析3种不同机器学习模型的预测精度,发现BP神经网络模型预测精度最高,其平均相对误差为8.55%。随后采用遗传算法优化BP神经网络。结果表明,所建立的BPGA模型平均相对误差比BP神经网络模型降低了6.47个百分点,其平均相对误差为2.08%,能够有效地为Cu-Ni-Co-Si合金固溶处理工艺参数的选择提供指导。The grain size of Cu-Ni-Co-Si alloy after solid solution treatment will affect its service performance.Three different machine learning methods,namely BP neural network,random forest,and long short-term memory network,were used to establish machine learning prediction models for the solid solution temperature and time of Cu-Ni-Co-Si alloy on the grain size after solid solution.The prediction accuracy of the three different machine learning models was compared,and it was found that the BP neural network model had the highest prediction accuracy,with an average relative error of 8.55%.Subsequently,genetic algorithm was used to optimize the BP neural network.The results showed that the average relative error of the established BP-GA model was reduced by 6.47%compared to the BP neural network model,with an average relative error of 2.08%,which can effectively guide the selection of process parameters for Cu-Ni-Co-Si alloy solid solution treatment.

关 键 词：Cu-Ni-Co-Si合金 固溶处理 工艺参数 晶粒尺寸 机器学习 

分 类 号：TG146.1[一般工业技术—材料科学与工程]