摘要
为实现化工过程稳态的快速准确检测,而开展基于统计分析方法的检测技术研究。主要从控制图法、CUSUM累积和法、多变量统计分析等统计方法概述,以及机器学习和深度学习等数据驱动法和模型驱动法对比分析。考察不同方法在检测精度、鲁棒性和计算效率方面的优劣,指出混合检测模型综合考虑先验知识与数据驱动的优点。最后,给出主成分分析判断炼油装置稳态的应用实例,为工业过程监测与控制提供参考。
In order to achieve rapid and accurate detection of steady-state chemical processes,research on detection techniques based on statistical analysis methods is carried out.This research primarily provides an overview of statistical methods such as control charts,CUSUM(Cu-mulative Sum Control Chart),and multivariate statistical analysis,as well as a comparative analysis of data-driven methods like machine learning and deep learning versus model-driven methods.The study examines the advantages and disadvantages of different methods in terms of detection ac-curacy,robustness,and computational efficiency,highlighting the benefits of hybrid detection models that integrate prior knowledge with data-driven approaches.Finally,an application example of using principal component analysis to determine the steady state of a refinery unit is provided,offer-ing a reference for industrial process monitoring and control.
作者
边尚芸
李彦鹏
杨林
Bian Shangyun;Li Yanpeng;Yang Lin(Beijing Jiutongqu Testing Technology Co.,Ltd.,Beijing,100000)
出处
《当代化工研究》
CAS
2024年第10期86-88,共3页
Modern Chemical Research
关键词
化工过程
稳态检测
统计分析
主成分分析
数据驱动
混合模型
chemical process
steady state detection
statistical analysis
principal component analysis
data driven
hybrid model
