Study On The Compositive Fault Diagnose Method By Independent Component Analysis And Support Vector Machine

The main task of Fault Diagnosis technique is trying to give a characteristic description of the failure when system meets trouble, using it to accomplish fault detection,advance information,separation & identification,and further making fault decision. This subject based on the common Independent Component Analysis (ICA) and Support Vector Machine (SVM) in fault diagnosis, associating the two methods to make further study of fault diagnosis technique of butadiene distillation column. There are five chapters in this paper and the following is the major coverage.ChapterⅠ, Introduction. This chapter first explains the origin and research meaning of the subject, and then establishes the main research content and key technique through the investigation and study on fault diagnosis technique home and abroad and the status of fault diagnosis technique of butadiene distillation column.ChapterⅡ, Accounting Theoretical Principle. This chapter introduces the ICA and SVM respectively, analyzing the sphere of application and characteristic, laid the theoretical foundation for the subsequent fault diagnosis control.ChapterⅢ, ICA-SVM Fault Diagnosis. This chapter based on the study of ICA and SVM, integrating the two together, and then brings the ICA-SVM model forward.ChapterⅣ, Simulation Testing. Associating the data produced in the procedure of fault diagnosis technique of butadiene distillation column, according to the built system model of fault diagnosis control, this chapter constructs the system model applied to fault diagnosis technique of butadiene distillation column, using the Matlab software, to do simulation testing on fault diagnosis technique of butadiene distillation column through research and simulation design. The result is helpful to compare the correlated parameter of the model optimize, and further lay the theoretical and testing foundation for the actual application of this method.ChapterⅤ, Summary & Outlook. This chapter summarizes the major coverage of the paper, giving a presentation of the insufficiency of the testing and making an expectation of the research in future.Creative viewpoints of this paper:⑴It's difficult to diagnose the original fault fast and accurately and wasting lots of sources by using ICA alone, and it's possible to make mistakes when classifying the problems by using SVM alone, here this paper presents the Fault Diagnosis technique integrating ICA-SVM.⑵This paper constructs the Fault Diagnosis System model of ICA-SVM,and explains the integrated step and procedure.⑶In Matlab simulation testing, ICA-SVM fault diagnosis system model is used to do experiment on DA106 and DA107 respectively. The result is helpful to compare the correlated parameter of the model optimize, and further lay the theoretical and testing foundation for the actual application of this method.