Skip to content Skip to main navigation Report an accessibility issue

Data-Enabled Anomaly Detection in Smart Manufacturing

Shenghan Guo, Ph.D.
Department of Industrial and Systems Engineering
Rutgers, the State University of New Jersey
Friday, January 22, 2021
3:30-4:20pm via Zoom


Advanced manufacturing, e.g., laser-based additive manufacturing (LBAM), is a pillar of Industry 4.0. However, the manufacturing process can be unstable and lead to part defects, thus limiting a wider adoption of the technique in industry. Recent development in inline sensing has enabled real-time information collection from manufacturing processes. The information comes in the form of time series data or in-situ thermal images and are valuable resources for early detection of process anomaly. My research explores statistical analysis and deep learning to develop data-enabled process modeling and monitoring approaches, which recognize defects or process deteriorations in real time by learning from the data. Three method components were developed to handle different data forms: (1) nonparametric trend detection in leak testing with time series, (2) in-situ detection of porosity in LBAM with thermal images, and (3) nondestructive quality prediction in resistance spot welding with thermal videos. The developed methods have been validated with real data and demonstrated for effectiveness in anomaly detection.


Dr. Shenghan Guo received the B.S. degree in Financial Engineering from Jilin University, Changchun, China, in 2013, the M.S. degree in Financial Mathematics from the Johns Hopkins University, Baltimore, MD, U.S., in 2014, and the M.S. degree in Engineering Sciences and Applied Mathematics from Northwestern University, Evanston, IL, U.S., in 2016. She joined the Ph.D. program in Industrial and Systems Engineering at Rutgers University, Piscataway, NJ, U.S. in 2016, and has successfully defended her Ph.D. in May 2020. Her research focuses on statistical process modeling and data-driven predictive analytics. The methods developed have been used to facilitate smart manufacturing, e.g., downtime reduction in powertrain manufacturing, defect prediction in laser-based additive manufacturing, nondestructive quality evaluation in resistance spot welding. She was awarded the Tayfur Altiok scholarship and the department nominee of Louis Bevier Dissertation Completion Fellowship in 2019. Outside the department, she was the finalist of Quality, Statistics, and Reliability (QSR) Paper Competition in 2018, finalist and winner of Quality Control and Reliability Engineering (QCRE) Data Challenge in 2019, and finalist and 2nd place of Data Analytics and Information Sciences (DAIS) Student Data Analytics Competition in 2020.