Multivariate statistical process control (MSPC) has been successfully applied to performance monitoring and fault diagnosis for chemical processes However, traditional MSPC are based upon the assumption that the separated latent variables must be subject to normal probability distribution, which sometimes can not be satisfied In this paper, a novel method combining principal component analysis (PCA) and independent component analysis (ICA) is proposed to model non Gaussian data from industry and improve the monitoring performance of process In order to deal with the uncertainty of probability distribution within the independent component, a kind of classifier referred to as support vector classifier is used for classifying the abnormal modes Simulation result for a nonisothermal continuous stirred tank reactor (CSTR) by the presented method verifies the effectiveness of ICA based
In industrial processes, measured data are often contaminated by noise, which causes poor performance of some techniques driven by data Wavelet transform is a useful tool to de noise the process information, but conventional transaction is directly employing wavelet transform to the measured variables, which will make the method less effective and more multifarious if there exists lots of process variables and collinear relationships In this paper, a novel multivariate statistical projection analysis (MSPA) based on data de noised with wavelet transform and blind signal analysis is presented, which can detect fault more quickly and improve the monitoring performance of the process The simulation results applying to a double effect evaporator verify higher effectiveness and better performance of the new MSPA than classical multivariate statistical process control(MSPC)
针对化工生产中日益增多的间歇过程,提出了一种基于多元统计信号处理的过程监控方法,其主要思想为将过程信息空间划分为由盲源信号描述的信号子空间、过程主元描述的信号子空间和残差信号子空间,随后对各个信号子空间构造过程统计量或分类器提取信号特征进行过程监控,该方法避免了传统多元统计过程控制(mult-ivariate statistical process contro,lMSPC)需假设过程特征信号服从正态分布的前提.将本方法与传统MSPC方法的性能进行了对比,并在仿真中给出了对比研究结果.通过对间歇过程的仿真研究表明,该方法不仅能够有效地检测出故障,而且有利于故障的分离和定位,从而说明该方法不仅是有效的,而且其性能优于仅能检测故障的传统MSPC过程监控方法.
