Study Of Industrial Process Monitoring With Independent Component Analysis

Independent component analysis (ICA), which is applied in the field of industrial process as a data analysis method, can extract the intrinsic factor of the process under the meanings of the statistical independence, and describe the essential characteristics of the production more efficiently. In this thesis the algorithm and application of ICA are further investigated for its application in process monitoring, then based on continuous stirred tank reactor (CSTR), a complete framework of statistical process monitoring with ICA is proposed, and an anticipated result is obtained through analysis and simulation according to the diverse instances that are disturbed from different variables.First, the thesis analyzes the principle and algorithm for ICA, and gives a new method of eliminating the untrue independent components with correlation coefficient for improving the accuracy. Second, in view of the approach and characteristics of multivariate statistical process control,...