With the insight of the current wave of data science and industrial internet revolution, Dr. Li successfully identify such topics will be a valuable discussion for the community of statistician. He then confirmed speakers, prepared the proposal, and finally won one of the four highly-influence well-attended Introductory Overview Lectures for the 2014 Joint Statistical Meetings. His speakers includes: Bill Ruh (VP of General Electric and Director of GE Software), Sokwoo Rhee (White House Presidential Innovation Fellow on Cyber-Physical Systems), and Michael Rappa (Director of Institute for Advanced Analytics, NCSU). Here is the link for the session detail.
Dr. Li is an expert on applying Bayesian analysis for reliability related problems where valuable solid prior information can be utilized especially when engineering and physical knowledge is available. He presented a webinar with the Reliability Division of ASQ with topic of "General Bayesian Methods for Typical Reliability Applications" with recorded video. An overview article was later published in Journal of Quality Technology in 2014. Dr. Li and Dr. Meeker were also invited to give a talk on this topic in the 58th Fall Technical Conference.
Bayesian methods becomes much more popular in reliability analysis. However the fact about from where the needed prior distribution should come is still a major concern. How to translate prior information into a prior distribution is another hurdle. Even with the prior distribution ready, the complicated computation may prevent reliability engineers to implement the Bayesian methodology. With a few consultation projects in the area of Bayesian reliability, a set of systematic analytic procedures have been introduced by Dr. Li and Dr. Meeker. A detailed discussion with a few classical reliability problems using Bayesian approach was published on the January 2014 issue of Journal of Quality Technology.
Aircraft safety is essential for both passenger and cargo transportations. One way to ensure the safety is through periodic inspections and maintenance through various nondestructive evaluation (NDE) methods. Vibrothermography and Eddy-Current are two widely used in aircraft safety NDE. There are all sorts of variations involved in these measurement procedures and statistical uncertainty needs to be quantified during the data analysis. Dr. Li, Dr. Meeker, and other team members systematically invested the statistical methodology behinds NDE data analysis. The results are summarized in two Federal Aviation Administration documents: Ultrasonic Probability of Detection Curves for Synthetic Hard Alpha Inclusions in Titanium Forgings (pdf) and Thermal Acoustic Studies of Engine Disk Materials (pdf).
Hard alpha inclusions in titanium alloy aircraft engine disks can lead to serious accidents (i.e. NTSB/AAR-90/06). A physics-model assisted statistical method is introduced to analyze the multi-site measurement data obtained from a synthetic inclusion forging disk. The proposed method enables needed information extraction from data taken on the limited types and sizes of the synthetic inclusions and further makes possible interpolations and extrapolations for a wide range of size and concentration. This work was published on the February 2014 issue of Technometrics.
Another example of a typical consulting project resulted a journal publication (ASMBI) with newly development statistical methodology in reliability, availability and maintenance. An analytical framework is developed for estimating system unavailability metric based on historical data of a fleet of heavy-duty industry equipment. During the useful life of such system, repairs and maintenance actions are performed. However, not all repairs or maintenance actions were recorded. Specifically, the information on event times, types, and durations is available only for certain time intervals (i.e., observation windows), and special statistical consideration is needed to adjust the observation window effect.
Regular aircraft inspection and maintenance are crucial to ensure air travel safety, and nondestructive evolution methods are usually used to detect flaws or cracks for aircraft parts such as engine fan blades, and fuselage. Currently there are industry standards for scheduling, measuring, recording and analyzing inspections. With the development of sensor technology, information storage and extraction, and analytical powers, it becomes feasible to use much more data than the standard procedures to return more accurate and robust results. In this project, a modern data acquisition procedure and advanced analytics framework are introduced to integrate sensor, data and analytics. The results were published in 2012 Issue of Research in Nondestructive Evaluation.