Fault Detection Of Multi-stage Fermentation Process Analysis Based On Kernel Entropy Component

Fermentation process is a typical batch process,its application is in the field of food,medicine and other industries with good prospects for development.To ensure safe and efficient production is very important to the fermentation industry.Aiming at the nonlinear and multi-stage characteristics,a multi-stage fermentation process monitoring algorithm based on kernel entropy component analysis is proposed for stage divison and fault monitoring the fermentation process for the fermentation process.(1)Research on constructing a statistic based on angle structureIn view of the complex and nonlinear characteristics of fermentation process,a method of multi-stage kernel entropy component analysis for fermentation process monitoring based on angle structure statistics was studied.This method uses KECA to extract the principal component matrices of fermentation process data.The results show that the principal component matrices has a good angular structure characteristic,so the principal component matrices are used to construct the statistics based on the angular structure.Compared with the traditional statistics,it is not necessary to assume that the process variables obey the Gauss distribution.(2)Stages divison based on FCM clustering algorithmAccording to the multi-stage characteristic of fermentation process,the fuzzy Cmeans(FCM)clustering algorithm is studied to realize the division of fermentation process.The fermentation process is divided into several stages,which can improve the sensitivity of fault detection.In addition,In addition,in view of the unequal length of the fermentation process,a method based on similarity judgment theory is proposed to realize the ownership judgment of the sampling points in on-line monitoring.(3)Research on a fault detection method based on kernel entropy component analysisBased on the multi-stage and nonlinear characteristics of fermentation process,this paper presents a new method of multi-stage fermentation process monitoring based on kernel entropy component analysis.Firstly,the FCM clustering algorithm is used to divide the fermentation process into several sub stages,and then the MKECA fault detection model is established for each sub stage.The method can detect the fault rapidly,and the false alarm rate is significantly reduced.(4)Experiment of Escherichia coli fermentationFinally,the research method is applied to the experiment of E.coli fermentation.The experimental results show that the method can effectively extract the data of the production process,and quickly detect the occurrence of the fault.Compared with the MKPCA method,the false alarm rate is significantly reduced and the monitoring sensitivity is higher.