A Visual Method Of Fault Diagnosis Based On Self-organizing Map

With the increasing complexity of industrial process, fault diagnosis has attracted much attention recently. Visualization has been used routinely in data mining as a presentation tool to generate initial views, navigate data with complicated structures, and convey the results of an analysis. A new visual fault diagnosis method based on self-organizing map for complicated industrial process is researched in this thesis. How to deduce the cause of the certain fault is also discussed.The main contributions can be summarized as follows:1. The development and applications of fault diagnosis methods are reviewed, as well as the characteristics of fault and fault detection for complicated process are analyzed. The application of visualization in fault diagnosis is briefly introduced.2. The connect form and the study method of artificial neural network are introduced, the characteristic and the achieve method of Self-organizing map are analysed, besides, its application for nowadays is summarized.3. A new fault diagnosis method is proposed based on SOM. Principal Component Analysis and wavelet are utilized for the preprocessing of the data set. The cause of certain fault can be deduced from the U-matrix of the derived SOM network and the loadings vector of the principal components. Online monitoring can be achieved by the trajectory in the visualization space.4. The application of the proposed method on the Tennessee Eastman process is studied. The results illustrate the effectiveness of the method.