摘要
为了解决传统PCA方法应用于含有噪声干扰数据时产生较高误报率和漏报率的问题,提出一种基于函数型数据分析的暖通空调系统故障检测与诊断方法。首先,将离散的测量数据看作一个完整的函数对象,通过基函数的线性组合给出函数估计,消除异常值补全缺失值;其次,将估计的函数离散化作为训练和测试数据矩阵;最后,针对传感器偏差和漂移两种故障进行仿真研究。结果表明,与传统PCA和中值滤波方法相比,改进的PCA方法不仅可以滤除噪声和剔除异常值,而且降低了故障误报率和漏报率。
In order to solve the problem that the traditional PCA method generates relatively high false alarm rates and false negative rates when it is applied to data with noise,a novel fault detection and diagnosis method based on functional data analysis is proposed for HVAC system.Firstly,the discrete measurement data is regarded as a complete function object,the function estimation is given through the linear combination of the basis function,missing values are completed and outliers are eliminated;Secondly,the estimated function is discretized as the training and the testing matrix.Finally,using simulation experiments to test the performance of the algorithm on two kinds of faults:bias fault and drift fault.Compared with the traditional PCA method and median filter PCA,the improved method not only can effectively filter out noise and eliminate the abnormal value,but also can reduce the false alarm rates and false negative rates.
作者
单彪
堵俊
商亮亮
SHAN Biao;DU Jun;SHANG Liang-liang(School of Electrical Engineering,Nantong University,Nantong,Jiangsu 226019,China;Shanghai Micro Electronics Equipment(Group)Co.,Ltd,Nantong,Jiangsu 226019,China)
出处
《控制工程》
CSCD
北大核心
2020年第4期765-770,共6页
Control Engineering of China
基金
国家自然科学基金面上项目(61973176)
江苏省高等学校自然科学研究项目(17KJB120010,18KJB12007)。
关键词
暖通空调
故障检测
函数型数据分析
主成分分析
Heating ventilating and air conditioning
fault detection
functional data analysis
principal component analysis
