Building and updating Independent Component Analysis model in dual rate system

摘要

Independent Component Analysis (ICA) is a simple,powerful tool for analyzing multi-variant data to find abnormal on process monitoring.Traditional ICA algorithm is implemented on the consumption of only one sampling rate in the control system.However,there are a few multi rate control systems that could not apply the primitive ICA algorithm to the surveillance of industrial process.Several approaches of multivariate statistical modelling ICA are presented on dual rate system whose variables are correlated each other in this paper.Furthermore,a strategy of updating ICA model on the dual rate system is also suggested with restricted number of new samples to replace the equally number of earliest one.The simulation results verify the capability of the modelling and updating ICA on dual rate system with their calculated statistics.