BEGIN:VCALENDAR BEGIN:VEVENT SUMMARY:PhD Defense DESCRIPTION:Speaker: Yuan Jin , Mork Family Department of Chemical Engineering Talk Title: Statistical modeling and process data analytics for smart manufacturing Abstract: Smart manufacturing has been the focus of many researchers and has been extended to varies areas. With the increased complexity of the manufacturing process and the variety data collected from various aspects throughout the process, process data analytics is thus essential to discover process knowledge and predict the future production. In this dissertation, we introduced two challenges accompanied by smart manufacturing and discussed how we handle the challenges for processes with different manufacturing characters. The two challenges are "complexity" and "variety", the areas of applications are additive manufacturing (AM) and pharmaceutical manufacturing.\n \n To handle the challenges accompanying with AM, we first study a statistical modeling and optimal compensation approach to predict and improve the shape accuracy of AM printed parts, especially for the out-of-plane deviation. This method is data driven and thus, not blocked by the complicated AM physical mechanism. Moreover, this method is able to deal with low volume sample data and high volume geometric variety. The feasibility and effectiveness of this approach is proved by experimental study. \n \n To deal with the challenges accompanying with pharmaceutical manufacturing, we proposed a two-stage strategy to study a large-scale cell culture manufacturing process variability. This strategy not only adopts multivariate analysis (MVA) and machine learning (ML) methods on intricate multiple-step bio-processes, but also takes use of multilevel heterogeneous datasets to unveil hidden process characteristics and provide insights into factors affecting process quality. This strategy has been applied to a real antibody pharmaceutical manufacturing, pointing to new cues for domain experts to better understand the process.\n Biography: Yuan Jin is a Ph.D. candidate in Mork Family Department of Chemical Engineering & Materials Science at University of Southern California (USC), Los Angeles, CA. She received her B.E. degree in Electrical Engineering from Zhejiang University, Hangzhou, Zhejiang in 2013. Her research is about statistical modeling and process data analytics for smart manufacturing. Host: Dr. Joe Qin DTSTART:20181002T130000 LOCATION:HED 116 URL;VALUE=URI: DTEND:20181002T150000 END:VEVENT END:VCALENDAR