基于python增量学习的薄膜在线测长仪智能校准系统  

Intelligent calibration system of thin film online length measuring instrument based on python incremental learning

作　　者：张乐 刘小谜 庄苏宁 于辉 ZHANG Le;LIU Xiaomi;ZHUANG Suning;YU Hui(Suqian Institute of Metrology and Testing,Suqian 223800,China)

机构地区：[1]宿迁市计量测试所,江苏宿迁223800

出　　处：《工业计量》2025年第1期59-63,共5页Industrial Metrology

基　　金：江苏省市场监管局科技计划项目(KJ2024100)。

摘　　要：针对当前薄膜测长仪无法校准的难题,研究利用增量学习技术,创新设计了智能校准装置和软件。装置能自动采集、记录和校准传感器数据,构建数学映射模型,实现测量误差的在线智能校准。通过定制化的校准软件,实时读取分析传感器参数,提高了校准效率和准确性。同时,引入拉依达检验法剔除异常值,蒙特卡洛法进行不确定度评定,梯度下降法进行相关性预测,进一步提升了校准的可靠性和有效性。该研究将为膜材料产业的测长仪发展提供计量技术参考。Aiming at the problem that the current thin film length measuring instrument cannot be calibrated,this research uses incremental learning technology to innovate and design an intelligent calibration device and software.The device can automatically collect,record and calibrate sensor data,build a mathematical mapping model,and realize online intelligent calibration of measurement errors.Through customized calibration software,real-time reading and analysis of sensor parameters improves the cali-bration efficiency and accuracy.At the same time,the introduction of Laida test method to remove outli-ers,Monte Carlo method for uncertainty assessment,and layer descent method for correlation prediction further enhances the reliability and effectiveness of calibration.This research will provide a metrological technical reference for the development of length measuring instruments in the membrane material industry.

关 键 词：增量学习 在线薄膜测长仪 异常值剔除 不确定度评定 相关性预测 

分 类 号：TH711[机械工程—测试计量技术及仪器]