摘要
针对主元分析(Principal component analysis,PCA)和局部保持投影(Locality preserving projections,LPP)方法在降维过程中分别只能保留数据集的整体信息和局部信息,提出一种基于局部整体结构保持投影的贝叶斯故障检测与辨识方法(Local and global structure preserving projections and bayes,LGSPP-Bayes);首先,将正常工况操作下的原始数据通过局部整体结构保持投影方法投影到低维特征空间,得到高维到低维的数据转换矩阵;然后通过设计贝叶斯分类器来进行故障检测;最后当检测到故障后通过计算贝叶斯分类函数的大小来识别故障种类;将LGSPP-Bayes方法应用于TE过程,仿真结果表明对故障的检测优于其他方法,并且可以很好地将故障种类识别出来。
According to the method of principal component analysis（PCA）and locality preserving projections（LPP）can respectly retain the global information and local information of the data set in the process of reducing dimension,a novel method named local and global structure preserving projections and bayes（LGSPP-Bayes）was proposed.Firstly,projecting the original data under normal operating conditions onto a low dimensional feature space to get a data transformation matrix from high dimension to low dimension;Then designing bayesian classifier for fault detection;Finally when a fault was detected,identifying which kind of the fault is by calculating bayesian classification functions.Case applying to Tennessee Eastman process illustrates the new method is better than other methods in the detection of fault.Besides,it is also a good way to identify fault types.
出处
《计算机测量与控制》
2015年第7期2288-2291,共4页
Computer Measurement &Control
基金
特殊环境机器人技术四川省重点实验室开放基金(13zxtk06)
西南科技大学博士基金
西南科技大学研究生创新基金(15ycx118)
关键词
主元分析
局部保持投影
贝叶斯分类器
故障检测
故障辨识
principal component analysis（PCA）
locality preserving projections（LPP）
Bayesian classifier
fault detection
fault identification
