Research On Fault Detection Method Of Beer Fermentation Process Based On ICA-SVM

Process monitoring plays a very important role in modern industrial production.The effect of the process monitoring directly affects the efficiency of industrial production.Traditional and statistical process monitoring is a data-driven method based on multivariate statistics.By monitoring the process,processing the monitored data,it can get the condition under normal working conditions,and it can also appear in the process of detection and diagnosis of the abnormal circumstances,according to the abnormal situation to deal with to ensure the operation of the entire system to ensure the production efficiency.In this paper,combined with the typical intermittent chemical production process,that is,the beer fermentation process as the object of study,because ICA can not accurately isolate the fault,so the ICA alone applied to the beer fermentation process can not accurately diagnose the specific causes of failure.The SVM classifier can isolate the fault,but it is easy to cause the wrong problem.In view of this,this paper presents an ICA-SVM integrated fault diagnosis method,and its application to the beer fermentation process.The main work is:1.First introduced the beer fermentation process need to monitor the variables and the need for fault diagnosis,and the introduction of ICA-SVM model.2.The independent component analysis algorithm and the estimation principle are introduced,and the shortcomings of ICA alone are discussed.It is found that ICA is difficult to isolate fault accurately.And the SVM can classify the fault data,so the introduction of SVM,introduced the advantages and disadvantages of SVM.3.As a result of the use of SVM alone is likely to cause the problem of misclassification,this chapter proposes an ICA-SVM integration algorithm,selection the data of the fault samples and the normal samples from the beer of experiment in the past few years to simulate the experiment.4.In view of some key variables of the beer production process,for example,temperature,pressure and other factors influence on the beer fermentation process,the ICA-SVM fault diagnosis model is used in the beer fermentation process,through the experimental data to verify the feasibility of ICA-SVM.