摘要
为了提升常量金标准物质标准值的不确定度评定精度,文章采用传统GUM方法与机器学习方法进行比较,研究不确定度的来源及影响因素。通过对电阻测量数据的分析,探讨了两种方法在多参数耦合与非线性影响下的表现与适用性。结果表明,机器学习方法能够在数据驱动的基础上提高预测精度,而传统方法则在理论推导上更为严谨。文章为高精度计量体系中的不确定度评定提供了新的思路与方法。
In order to improve the accuracy of uncertainty assessment of the standard values of constant gold standard materials,traditional GUM method and machine learning method were compared to study the sources and influencing factors of uncertainty.Through the analysis of resistance measurement data,the performance and applicability of two methods under multi parameter coupling and nonlinear effects were explored.The results indicate that machine learning methods can improve prediction accuracy on a data-driven basis,while traditional methods are more rigorous in theoretical derivation.This paper provides new ideas and methods for uncertainty assessment in high-precision metrology systems.
作者
车婧
王晓晓
曲鹏冲
马浩然
宫汝杰
CHE Jing;WANG Xiaoxiao;QU Pengchong;MA Haoran;GONG Rujie(China Geological Survey Yantai Coastal Zone Geological Survey Center,Yantai 264000,China)
出处
《化工管理》
2025年第31期45-48,共4页
Chemical Management
关键词
常量金标准物质
不确定度评定
传统方法
机器学习
误差传播
constant gold standard substance
uncertainty assessment
traditional methods
machine learning
error propagation
