Track: Doctoral Dissertation Competition

Abstract

Due to the advancement of modern industrial processes, multivariate statistical analysis methods have been developed progressively to handle complex systems by exploiting valuable information from the collected data for predictive modeling and fault diagnosis, such as partial least squares, canonical correlation analysis, latent variable regression and their extensions. However, these methods suffer from issues involving comprehensive monitoring and dynamics. To address them, a concurrent kernel LVR (CKLVR) is designed for collinear and nonlinear data to construct a full decomposition of the original nonlinear data space, and to provide comprehensive information of the systems. A dynamic auto-regressive LVR (DALVR) is also proposed based on regularized LVR to capture dynamic variations in both process and quality data. The concurrent monitoring and fault diagnosis and causal analysis scheme based on DALVR are also developed. Their superiority can be demonstrated with the Tennessee Eastman Process case study.