基于定性趋势分析的空调系统传感器故障检测  

A Detection Method Based on Qualitative Trend Analysis for Sensor Faults in HVAC Systems

作　　者：宋政辉[1] 应洪伟[1] 潘晓[1] Song Zhenghui;Ying Hongwei;Pan Xiao(Xichang Satellite Launch Center,Xichang 615000,China)

机构地区：[1]中国西昌卫星发射中心,四川西昌615000

出　　处：《计算机测量与控制》2018年第11期30-34,共5页Computer Measurement &Control

摘　　要：随着DCS、各种智能化仪表、现场总线技术在空调系统中的广泛应用,大量的数据被采集并存储下来,但是这些包含空调系统传感器状态信息的数据并没有得到有效的利用;文章提出了基于定性趋势分析的数据处理和分析方法,挖掘出数据的隐含信息,以检测空调系统传感器故障;该方法将空调系统中信号相似的传感器分成一组,利用组中信号间的趋势相似性进行故障检测;采集了空调系统的传感器数据,对传感器偏置故障和漂移故障进行仿真实验,结果表明,该方法能检测传感器的偏置故障和漂移故障;但该方法计算量会随着同一组中传感器数量的增加而呈几何级数增加,故传感器分组时应当控制同一组中传感器的个数。With the application of DCS,various intelligent instruments and fieldbus technology in HVAC system,a large number of data are collected and stored,but these data containing the state information of HAVC system sensors have not been effectively utilized.In this paper,a method of data processing and analysis based on qualitative trend analysis is proposed to mine the hidden information in the data to detect sensor faults in HVAC systems.It divides sensors which have similar signals into groups firstly,then uses the trend similarity of signals in one group for fault detection.By collecting sensor data of air conditioning system,the bias fault and drift fault of the sensors are simulated.The results show that the method is feasible.The computational complexity of this method will increase with the increase of the number of sensors in the same group,so the number of sensors in the same group should be controlled when the sensors are grouped.

关 键 词：定性趋势分析 传感器 故障检测 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]