报告题目: Towards Big Data Process Analytics and Monitoring
报告人: Prof. S. Joe Qin （秦泗钊教授）, University of Southern California
时间: 2013年5月18日(周六)下午2:00
地点: 实验15楼207
报告人简介:
S. Joe Qin, IEEE Fellow, is the Fluor Professor of Process Engineering and Vice Dean at the Viterbi School of Engineering at University of Southern California. He obtained his B.S. and M.S. degrees in Automatic Control from Tsinghua University in 1984 and 1987, respectively, and Ph.D. degree in Chemical Engineering from University of Maryland at College Park in 1992. His research interests include big data process analytics, statistical process monitoring, fault diagnosis, model predictive control, system identification, run-to-run control, and control performance monitoring. He is a Co-Director of the Texas-Wisconsin-California Control Consortium which celebrates the 20th anniversary this year. He is a recipient of the National Science Foundation CAREER Award, the 2011 Northrop Grumman Best Teaching award at Viterbi School of Engineering, the DuPont Young Professor Award, Halliburton/Brown & Root Young Faculty Excellence Award, and an IFAC Best Paper Prize for a paper published in Control Engineering Practice. He is currently an Associate Editor for Journal of Process Control and a Member of the Editorial Board for Journal of Chemometrics. He served as an Editor for Control Engineering Practice and an Associate Editor for IEEE Transactions on Control Systems Technology. He has published over 100 archival journal papers with an ISI Web of Science h-index of 32.
Abstract
Process monitoring provides supervision of process operations so that abnormal operating conditions can be detected, diagnosed, and proper adjustment can be implemented as needed. The focus of this talk is on the recently rising interest of big data of multiple sources and the use of multivariate statistical methods for multi-level data-driven process monitoring. The ultimate purpose is to reduce process variability under real-world operating conditions with the use of real time data.
A concurrent projection to latent structures is presented for the monitoring of output-level faults that affect the quality and input-level process faults. The input and output data spaces are concurrently projected to five subspaces, a joint input-output subspace that captures co-variations between input and output, an output-principal subspace, an output-residual subspace, an input principal subspace, and an input-residual subspace. Statistical indices are developed for various fault detection alarms based on these subspaces. The proposed monitoring method offers complete monitoring of faults that happen in the predictable output subspace and the unpredictable output residual subspace, as well as faults that affect the input spaces only.