EAPSI:Investigating a Nonparametric Method to Predict Longevity of Engineering Monitoring Systems

Schuette Geoffrey H, Arlington TX

Investigators

Abstract

As systems in engineering continue to become larger and more complex in nature, real time condition monitoring of such systems continues to be an important area of study in order to more accurately predict system failure and develop systems to prevent catastrophic consequences. Using probability and statistics tools, this research will develop a robust method to predict long term conditions of the systems being analyzed and to do so under conditions of unknown prior distributions. The research will be conducted under the mentorship of Professor Ming-Yen Cheng of National Taiwan University. In the Bayesian approach, prior distributions are assumed, and posterior distributions are then established from assumptions of the prior and the model. Hence, we can find the probability via the appropriate threshold failure time and access the long-term condition of the system. When the prior is unknown, however, it is then important to be able to estimate the prior from samples generated by the signal time. The focus of this research proposal is to fit the unknown prior distribution with a smooth empirical distribution function based on kernel function and sequence of bandwidths to estimate the unknown prior distribution. The research will be conducted under the mentorship of Professor Ming-Yen Cheng, a noted statistician of National Taiwan University Department of Mathematics. This award under the East Asia and Pacific Summer Institutes program supports summer research by a U.S. graduate student and is jointly funded by NSF and the Ministry of Science and Technology of Taiwan.