Research On The Improvement Of Fault Monitoring Method In Multiphase Batch Process

With the development of science and advancement of technology,the process monitoring of industrial production becomes the research hotspot in the field.Batch process is an important industrial production method.The process mechanism is complex and there are multiple operation stages,and the product quality is easily affected by uncertainties.In order to reduce the loss of industrial production caused by process failure,it is urgent to establish an effective process monitoring system to monitor the failure of the batch process.This topic analyzes the research status of the batch process monitoring method,and combines the production characteristics and data characteristics of the batch process to conduct the research and improvement of the batch process monitoring method.The following is the main research content of this article:(1)This paper introduces the principal component analysis(PCA)applied to multivariate statistical process monitoring,and multi-way PCA(MPCA)method suitable for batch process monitoring is analyzed.Combining the characteristics of the batch process,two different expansion modes of MPCA method and data filling methods for online monitoring are discussed.The monitoring framework and monitoring steps of the MPCA method in the batch process are described.(2)Considering that online fault diagnosis of batch process involves the prediction of unknown output variables,a fault diagnosis approach based on data expansion and fault classifier data selection is therefore proposed in this paper.The three-dimensional dataset which contains batch information is unfolded to establish MPCA models for fault detection;A few continuous time samples after the fault occurred moment are selected for fault feature extraction,and a least squares support vector machine(LSSVM)fault classification model is built during the offline stage afterwards;Then,online fault diagnosis is realized through the established fault classifier,and classification and identification of fault can be achieved.It was applied to the simulation experiment of penicillin fermentation process.The results show that this method improves the real-time and accuracy of online fault diagnosis of batch process.(3)In view of the characteristics of multiple stages and the non-Gaussian of batch process,an improved stage division and fault monitoring method is proposed.Firstly,the stage is divided according to the similarity of each time slice and k-means algorithm,and then the independent component analysis(ICA)method is used to extract the feature information of non-Gaussian of each stage respectively.Finally,the support vector data description(SVDD)algorithm is introduced to establish a statistical analysis model for the independent components and the remaining Gaussian residual spaces,and the whole process is monitored.The feasibility and effectiveness of the proposed method is verified by an actual fault monitoring application for the semiconductor etch process.