摘要
冷水机组安全运行和优化节能的必要条件之一,是传感器测量数据能真实反应系统的运行工况。但随着使用年限的增加,传感器的小幅度故障极易发生且难以识别。本文在基于数据驱动的传感器故障检测、诊断与数据重构相关方法中,对比分析了基于主元分析和基于多重线性回归的传感器错误数据重构算法,并采用实际工程数据进行验证。结果表明,多重线性回归方法比主元分析方法精度更高,误差更小。
One of the essential conditions for the safety operation and optimal conserva-tion of chillers is that the data measured by the in-site sensors is reliable. Unfortunately,the low-level sensor faults are occurred easily and identified hardly due to the long term operation period. There are some data-driven methods have been employed for the sensor fault detection, diagnosis,and reconstruction. Among them, two reconstruction meth-ods ？ one based on the Principal Component Analysis and the other based on the Multiple Linear Regression, are investigated by the in-site data. Results show that the accurate de-gree of the Multiple Linear Regression is higher than that of the Principal Component A -nalysis,and the estimated error of the Multiple Linear Regression is lower than that of the Principal Component Analysis.
出处
《制冷与空调》
2017年第5期15-19,共5页
Refrigeration and Air-Conditioning
基金
2016年湖北省教育厅科学技术研究项目(B2016361)
武汉市科技局科技创新平台建设计划(2015061705011607)
武汉市教育科学“十三五”规划2016年度重点(专项)课题(课题批准号:2016A125)成果
2016年度武汉商学院校级教学研究项目(2016Y010)
关键词
冷水机组
传感器故障
主元分析
多重线性回归
数据重构
chiller
sensor fault
principal component analysis
multiple linear regre-sion
data reconstruction
