Statistical Processes Monitoring Based on Improved ICA and SVDD

摘要

An industrial process often has a large number of measured variables, which are usually driven by fewer essential variables. An improved independent component analysis based on particle swarm optimization (PSO-ICA) is presented to extract these essential variables. Process faults can be detected more efficiently by monitoring the independent components. To monitor the non-Gaussian distributed independent components obtained by PSO-ICA, the one-class SVDD (Support Vector Data Description) is employed to find the separating boundary between the normal operational data and the rest of independent component feature space. The proposed approach is illustrated by the application to the Tennessee Eastman challenging process.