At the Automation and Digital Manufacturing Lab, our research focus is to improve product quality, reduce production cost, and promote manufacturing efficiency through developing and applying methodologies in a wide range of fields, including manufacturing, quality engineering, statistics, and machine learning.

Big Data Enabled Smart Manufacturing

With the rapid development of sensing and communication technologies, today’s manufacturing industry is marching towards a Big Data era and a new generation of digitalization and intelligence. The availability of big data provides us with a golden opportunity to understand, monitor, control, and eventually digitalize the manufacturing industry. One key problem lies in how to effectively and efficiently utilize multi-source big data to support smart decision-making at multiple levels, e.g., machine, station, factory, and supply chain. We are currently developing methods that are are flexible and effective in modeling data with an extremely complex structure.

Multi-Level Integration of Modeling, Monitoring, and Control

Traditional manufacturing research and practice mostly consider the problems of modeling, monitoring, and control in isolation, thus leading to a lack of communication and overall optimization. By removing the barriers, we can potentially improve the effectiveness to a great extent. However, to derive a complete solution to this problem is highly challenging and requires a great deal of inter-disciplinary efforts. We are working on this problem in a “local-to-global” fashion, namely, we address the integration problem in individual levels first and then incorporate them globally.

Ultrasonic Metal Welder as a Testbed

Real-world manufacturing processes and data are crucial for validating the developed algorithms and methods. We are establishing a testbed based on ultrasonic metal welding process. Ultrasonic metal welding is a solid-state joining method that has been used in a variety of applications, e.g., battery joining in the production of GM’s Chevy Volt. Various sensors, such as power, force, LVDT, acoustic sensors, and 3D microscope will be used to collect data reflecting the process physics.

Data-Fusion for Health Care

Though hardly to be directly related, health care and manufacturing have similar data structures. Therefore, some data-fusion techniques could be shared by two fields. We are extending our work to health care with the hope that better utilization of multi-source data will assist with the diagnosis and treatment in health care.

As an example, our collaboration with the University of Texas MD Anderson Cancer Center has resulted in one paper published in Leukemia (impact factor: 12.104, ranking #1 in Hematology). Some results are shown by the figures below. See the publication here.