Speaker

 Chen, Yong， Associate Professor, University of Iowa

　　Title

     Reliability Analysis for Linear Sensor Systems             

　　Time

　　Venue

　　S712 

　　Abstract

A linear sensor system is defined as a sensor system in which the sensor measurements have a linear relationship to the source variables that cannot be directly measured. When considering catastrophic sensor failures, the reliability of a linear sensor system is defined based on its diagnosability performance. A mathematical tool called matroid theory is applied to study the reliability of the linear sensor system; properties of the minimal paths and minimal cuts are derived; and efficient methods are developed to evaluate the exact system reliability for two special types of systems. To overcome the computational complexity in evaluating reliability of general linear sensor systems, Monte Carlo methods are developed to approximate the sensor system reliability. The crude Monte Carlo method can be straightforwardly applied for the linear sensor system. But it is not efficient when the sensor system is highly reliable. An improved Monte Carlo method for network reliability, named Recursive Variance Reduction (RVR) method, is adapted for the reliability problem of linear sensor systems. In order to apply the RVR method, methods based on matroid theory are proposed to obtain minimal cut sets of the linear sensor system, particularly under the condition that the states of some sensors are fixed as failed or working. A case study is conducted for a linear sensor system used for fault diagnosis in multistage automotive assembly processes, which demonstrates the efficiency of the developed methods for reliability evaluation of linear sensor systems.

　　Affiliation

 Yong Chen is currently an associate professor in the Industrial Engineering program at the University of Iowa. He received the B. E. degree in computer science from Tsinghua University, China in 1998, the Master degree in Statistics and Ph. D. degree in Industrial & Operations Engineering, both from the University of Michigan in 2003. His research interests include fault-tolerance and reliability issues in complex systems, preventive maintenance decision making, statistical modeling of repairable systems, and statistical monitoring, diagnosis, and prognosis. He received Best Paper Awards from IIE Transactions in 2004 and 2010. He is a member of the INFORMS, the ASA, and the IMS.