基于Stacking回归模型的超导材料临界温度预测方法研究  

Research on prediction method for critical temperature of superconducting materials based on stacking regression model

作　　者：何素贞[1] 杨冬平 HE Suzhen;YANG Dongping(School of Mechanical and Electrical Engineering,Putian University,Putian 351100,China;Zhijiang Lab Frontier Basic Research Center,Hangzhou 311101,China)

机构地区：[1]莆田学院机电与信息工程学院,福建莆田351100 [2]之江实验室前沿基础研究中心,浙江杭州311101

出　　处：《延边大学学报(自然科学版)》2025年第1期32-36,共5页Journal of Yanbian University(Natural Science Edition)

基　　金：国家自然科学基金面上项目(12175242).

摘　　要：为了更好地预测超导材料临界温度,提出了一种基于Stacking回归模型的超导材料临界温度预测方法.该方法利用化学元素周期表的序数和原子数百分比构建特征向量,然后通过结合Extra Trees、Ridge Regression和XGBoost 3种算法的Stacking回归模型对临界温度进行预测.研究结果表明:该方法可较好地预测超导材料临界温度(决定系数R^(2)为0.93),且优于KamH提出的方法.与采用复杂特征工程结合机器学习的方法相比,该方法在简便性和有效性方面也具有一定优势。研究结果为超导材料临界温度的智能预测提供了新思路.To better predict the critical temperature of superconducting materials,a novel prediction method based on the Stacking regression model is proposed.This method constructs feature vectors using the atomic number and atomic percentage of elements from the periodic table and employs a Stacking regression model that combines Extra Trees,Ridge Regression,and XGBoost algorithms for prediction.The results show that this method achieves a high prediction accuracy for the critical temperature of superconducting materials(with a coefficient of determination R^(2) of 0.93),outperforming the method proposed by Kam H.Compared with methods that integrate complex feature engineering with machine learning,this approach offers certain advantages in terms of simplicity and effectiveness.The findings provide a new perspective for the intelligent prediction of the critical temperature of superconducting materials.

关 键 词：Stacking模型 超导材料 临界温度 机器学习 评价指标 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]