【摘要】 针对基于传统的多向主元分析(Multiway Principal Component Analysis,MPCA)方法用于间歇过程在线监控时需要对新批次未反应完的数据进行预估,从而易导致误诊断,且统计量控制限的确定是以主元得分呈正态分布为假设前提的缺陷,结合Fisher判别分析(Fisher Discriminant Analysis,FDA)在数据分类及非参数统计方法核密度估计(Kernel Density Estimation,KDE)在计算概率密度函数方面的优势,提出了一种FDA-KDE的间歇过程监控方法。该方法首先利用FDA求取正常工况数据和故障数据的Fisher特征向量和判别向量,获得Fisher特征向量的相似度:然后在提出偏平均集成平方误差(Biased Mean Integrated Squared Error,BMISE)交叉验证法确定KDE的带宽从而获得相似度统计量控制限的基础上,利用已获得的数据测量值对过程进行监控,避免了基于MPCA方法对未来测量值的预估;最后采用基于Fisher判别向量权重的贡献图方法来进行故障诊断。通过对青霉素发酵间歇过程应用表明,所提出的方法比传统的MPCA方法能更及时地监测出过程异常情况,更准确地判断异常发生的原因。更多还原

【Abstract】 When multiway principal component analysis(MPCA) is used for on-line batch monitoring,the future behavior of each new batch must be inferred up to the end of the batch operation at each time and false alarms are produced easily,and in view of limitations of confirmation of statistical control limits which assumes that principal component scores are subjected to multivariate normal distribution,combining fisher discriminant analysis(FDA) which has the advantage in data classification and nonparametric statistics method of kernel density estimation(KDE) which has the advantage in computing probability density function,A FDA-KDE method for batch process on-line monitoring is proposed.Firstly,in order to get similarity degree about fisher eigenvector,FDA is used to obtain fisher eigenvector and discriminant vector of normal working conditions data and fault data;Secondly,on the basis of using biased mean integrated squared error(BMISE) cross validation method proposed to define KDE bandwidth,therefore,getting similarity degree statistical control limits,the proposed approach uses present data and overcomes pre-estimating the unknown part of process variable trajectory in MPCA.Finally,weight contribution plot about the fisher discriminant vector is used to perform fault diagnosis.Application results on a penicillin batch fermentation process demonstrate that,in comparison to the MPCA method,the proposed method is more accurate and efficient to detect and diagnose the malfunctions.更多还原