Please join the Department of Chemical Engineering on Monday, February 25, for a guest lecture by Dr. Qinqin Zhu, a research scientist from Facebook Inc., speaking on Concurrent Monitoring of Process and Quality Faults with Canonical Correlation Analysis.
Abstract
Multivariate statistical process monitoring methods based on process variables and quality variables have been widely used in industrial processes, including robotics, chemicals, microelectronics manufacturing and pharmaceutical processes. Among them, three basic multivariate statistical methods include principal component analysis (PCA), partial least squares (PLS) and canonical correlation analysis (CCA).
In this talk, Dr. Qinqin Zhu will discuss the pros and cons of these methods and their corresponding monitoring schemes. She will highlight her research on the concurrent modelling and monitoring schemes based on CCA, including concurrent CCA (CCCA) based modelling, nonlinearity and dynamics handling, and quality-relevant monitoring and diagnosis. CCCA specializes in exploiting the variance and covariance in the process-specific and quality-specific spaces, and it retains the CCA's efficiency in predicting the quality while exploiting the variance structure using subsequent principal component decompositions.
To go one step further, Dr. Qinqin Zhu will discuss latent variable regression (LVR) and dynamic-inner LVR (DiLVR). LVR aims to minimize the prediction error between the input scores and the output scores, and it focuses on both the input variance structure and the prediction efficiency, which overcomes the drawbacks of PLS and CCA.
Biographical
Sketch
Dr. Qinqin Zhu is a Research Scientist at Facebook Inc., working on developing large-scale machine learning and deep learning algorithms for the ads recommendation system. Prior to joining Facebook in 2017, she obtained her Ph.D. degree in the department of Chemical Engineering at the University of Southern California, with the supervision of Dr. Joe S. Qin.
Her research addresses theoretical challenges and problems of practical importance in the area of Process Systems Engineering. Leveraging the power of mathematical modelling and optimization, she strives to develop advanced multivariate statistical analysis algorithms that enhance decision making in complex engineering systems. In particular, her work focuses on solving problems of scientific and industrial interest in the following areas:
- Process data analytics
- Statistical process monitoring and fault diagnosis
- Big data and machine learning
- Optimization and control of process and energy systems