Tutorial Session

An entire day of tutorials on PHM was taught by leading experts in the field. The tutorial session was scheduled for Monday, September 26, 2011. All tutorials are free to all registered PHM 2011 attendees.

Tutorial 1: Machine Diagnostics using Advanced Signal Processing [download]
Bob Randall, University of New South Wales
Monday, 8:00am – 9:50am

To perform machine condition monitoring using vibration analysis one needs to interpret response measurements, which are a compound of forcing function and transfer function components. When a change occurs, it is important to be able to distinguish whether it is due to a change at the source or in the transmission path, since a change in condition could be indicated by either. Measurement signals are also a mixture of responses to a number of different sources, since machine vibrations in general are considered multiple input, multiple output (MIMO). Because most techniques assume a single input, it is advantageous to first decompose the signal into a number of single-input-multiple-output (SIMO) systems. This can be accomplished by a number of techniques involving blind source separation. Then, a first division is often carried out into discrete frequency and random components, for example in gearboxes, where it has been shown that gear signals are basically deterministic, whereas bearing signals are stochastic, because of a small amount of random slip. A number of techniques can be used to achieve this separation, each with its own pros and cons. The tutorial compares a number of separation algorithms, and illustrates their application to the diagnostics of gears, rolling element bearings and IC engines.

Bob Randall is a visiting Emeritus Professor in the School of Mechanical and Manufacturing Engineering at the University of New South Wales (UNSW), Sydney, Australia, which he joined as a Senior Lecturer in 1988. Prior to that, he worked for the Danish company Bruel & Kjaer for 17 years, after ten years experience in the chemical and rubber industries in Australia, Canada and Sweden. While at Bruel & Kjaer he developed a short course in machine condition monitoring that was given about 40 times in 20 countries, and authored the book “Frequency Analysis”. At UNSW he was promoted to Associate Professor in 1996 and to Professor in 2001. He has degrees in Mechanical Engineering and Arts (Mathematics, Swedish) from the Universities of Adelaide (1961) and Melbourne (1971), respectively. He is the invited author of chapters on vibration measurement and analysis in a number of handbooks and encyclopedias, and a member of the editorial boards of four journals including Mechanical Systems and Signal Processing and Trans. IMechE Part C. His book Vibration-based Condition Monitoring: Industrial, Aerospace and Automotive Applications was recently published by Wiley. He is the author of more than 190 papers in the fields of vibration analysis and machine diagnostics, and has successfully supervised fourteen PhD and three Masters projects in those areas. Since 1996, he has been Director of the DSTO (Defence Science and Technology Organisation) Centre of Expertise in Helicopter Structures and Diagnostics at UNSW.

Tutorial 2: Introduction to Prognostics [download]
N. Scott Clements, Lockheed Martin
Monday, 10:10am – 12noon

This tutorial will provide an introduction to the design, implementation, evaluation, and verification of prognostic algorithms. Participants will first be shown an overall architecture to understand the key role and potential benefits of prognostic algorithms in a PHM/Maintenance/Sustainment system. Next, the tutorial will provide a primer on data-driven and model-based prognostic techniques, including the pros and cons typically associated with the different approaches. Concepts such as run-to-failure data, feature extraction, reasoning techniques, data trending, failure thresholds, and remaining useful life will be discussed. Finally, the practical challenges of implementing, evaluating, and verifying prognostic algorithms in real-world applications will be discussed. Throughout the presentation, examples will be used to further illustrate the ideas presented.

Scott Clements is a PHM Systems Engineer with Lockheed Martin Aeronautics Company in Fort Worth, TX. His research involves physics of failure fault models and associated prognostic techniques. His research interests include PHM, data mining, verification techniques, and fault tolerant control systems. He received his bachelor’s degree from Mississippi State University in 1996 and his master’s and doctoral degrees from the Georgia Institute of Technology in 1998 and 2003, respectively.

Tutorial 3: Data Mining in PHM [download]
Gautam Biswas, Vanderbilt University
Monday, 10:10am – 12noon

It is generally recognized these days that it is potentially very beneficial to collect a multitude of data which are the basis for any PHM system. However, operators find themselves sometimes in a situation where the data are collected without an in-depth analysis. As a consequence there are large amounts of, potentially heterogeneous, data from online and offline sources from both individual systems as well as fleets that are likely to contain information related to the health of the system. Often times, information where the anomaly occurred is missing. Similarly, the signature itself may be unknown as well. In these situations, data mining and machine learning techniques can be employed to extract candidates that appear to be anomalous by some measure. This tutorial will explore the techniques and tools in the area of data mining and machine learning and illustrate how they can be used to tackle the issues outlined above.

Gautam Biswas received the Ph.D. degree in computer science from Michigan State University, East Lansing, MI. He is currently with Vanderbilt University, Nashville, TN, as a Professor of computer science and computer engineering with the Department of Electrical Engineering and Computer Science and a Senior Research Scientist with the Institute for Software Integrated Systems. He conducts research on intelligent systems with primary interests in hybrid modeling, simulation, and analysis of complex embedded systems, and their applications to diagnosis and fault-adaptive control. As part of this work, he has worked on fault-adaptive control of fuel transfer systems for aircraft and Advanced Life Support systems for NASA. He has also initiated new projects in distributed monitoring, diagnosis, and prognosis, and health management of complex systems. In other research projects, he is involved in developing simulation-based environments for learning and instruction, and planning and scheduling algorithms for distributed real-time environments. His research has been supported by funding from NASA, the National Science Foundation, the Defense Advanced Research Projects Agency, and the Office of Naval Research. Dr. Biswas is a Senior Member of the IEEE Computer Society, Association for Computing Machinery, Association for the Advancement of Artificial Intelligence, and Sigma Xi Research Society. He is an Associate Editor for the IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS – PART A. He has served on the program committee of a number of conferences.