论文部分内容阅读
针对间歇过程特点和基于多向主元分析(Multiway Principal Component Analysis,MPCA)的间歇过程监控方法的缺陷,利用核映射在处理非线性过程和Fisher判别分析(Fisher Discriminant Analysis,FDA)在故障诊断能力上的优势,提出了基于递推多模型的核多向Fisher判别式分析(Recursive Multi-model Kernel Multi-way FDA,RMKMFDA)的间歇过程监测与故障诊断方法。该方法采用多模型核多向Fisher判别分析(Multi-model Kernel Multi-way FDA,MKMFDA)非线性结构代替MPCA单模型线性化结构,并提出确定时滞变量的算法;一旦通过MKMFDA监测出某一新批次过程正常,则模型参考数据库就随之更新:在线监控时通过比较核Fisher特征向量之间的欧氏距离来实现,而最优核Fisher判别向量用来鉴别故障类型。该方法在实时监控新的批过程时,只需利用已收集到的数据信息,且在线递推地更新模型参考数据库,提高了间歇过程监控的准确性,克服了MPCA不能处理非线性过程和实时性问题。通过采用RMKMFDA与移动窗多向主元分析(Moving Window MPCA,MWMPCA)方法对青霉素分批补料发酵过程的实时监控,结果表明RMKMFDA比MWMPCA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。
In view of the shortcomings of intermittent process characteristics and intermittent process monitoring methods based on Multiway Principal Component Analysis (MPCA), nuclear mapping is used to deal with nonlinear processes and Fisher Discriminant Analysis (FDA) , A method of batch process monitoring and fault diagnosis based on Recursive Multi-model Kernel Multi-way FDA (RMKMFDA) is proposed. The method uses the multi-model Kernel Multi-way FDA (MKMFDA) nonlinear structure instead of the MPCA single model linearization structure and proposes an algorithm for determining the time-lag variables. Once the MKMFDA is used to detect a certain The new batch process is normal, then the model reference database will be updated: Online monitoring by comparing the Euclidean distance between the kernel Fisher eigenvector to achieve, and the optimal kernel Fisher discriminant vector used to identify the type of fault. When the method of real-time monitoring of the new batch process, it only needs to use the collected data and information and recursively update the model reference database to improve the accuracy of batch process monitoring. It overcomes the problem that MPCA can not handle nonlinear process and real-time Sexual issues. Real-time monitoring of penicillin fed-batch fermentation process by using RMKMFDA and MWMPCA method showed that RMKMFDA could monitor process abnormalities more timely than MWMPCA and judge more accurately The reason for the abnormality.