摘要
为满足实际工业过程中的生产需求,复杂的化工过程往往会包含多种运行模态,而且过程数据不再单一地服从高斯分布或非高斯分布.过程数据的多工况分布特性以及同一工况下数据分布的不确定性使得传统的多元统计方法无法得到满意结果.针对复杂化工过程中多工况以及复杂数据分布的问题,提出一种基于局部邻域标准化策略(Local Neighborhood Standardization,LNS)的故障检测方法.首先,运用局部邻域标准化策略对历史数据集进行预处理,并充分考虑到邻域密度,再通过局部密度因子(Local Density Factor,LDF)构造监控统计量,进而对工业过程数据进行在线故障检测,最后通过数值例子和Tennessee Eastman(TE)过程验证本文方法的有效性.
Complex chemical processes often have multiple operating modes to meet the changes of produc- tion condition. The actual industrial processes often contain multiple operating modes and the process data is no longer solely Gaussian or non-Gaussian distribution. The multimode characteristics and the uncertain- ty of data distribution within one single mode make the conventional multivariate statistical process moni- toring methods unsuitable for fault detection. To solve the problem of multiple operating modes and com- plex data distribution, this paper proposed a novel multimode data processing method called local neighbor- hood standardization （LNS） and local density factor. The LNS was used in data preprocessing and the local density factor was used as a monitoring statistic value. The validity and effectiveness of the proposed method were illustrated through a numerical example and the Tennessee Eastman process.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
2015年第6期868-875,883,共9页
Journal of Shanghai Jiaotong University
基金
国家科学自然基金资助项目(61374140)
关键词
多工况过程监控
局部密度因子
局部邻域标准化
multimode process monitoring
local density factor
local neighborhood standardization
